# Connectors All documentation for Connectors. --- ## about-integrations-using-remote-mcp-2493848bcf.md # Get started with custom connectors using remote MCP Custom connectors using remote MCP are available on Claude, Cowork, and Claude Desktop for users on Free, Pro, Max, Team, and Enterprise plans. Free users are limited to one custom connector. ## What are custom connectors? Custom connectors let you connect Claude directly to the tools and data sources that matter most to your workflows. This enables Claude to operate within your favorite software and draw insights from the complete context of your external tools. You can: - Connect Claude to existing remote MCP servers. - Build your own remote MCP servers to connect with any tool. **Security and privacy with custom connectors** Custom connectors allow you to connect Claude to services that haven't been verified by Anthropic. Once connected, Claude can access those services and take action in them. For more guidance, review the **[Security and privacy considerations](#h_b79c05dfcd)** section below. ## What are remote MCP servers? The Model Context Protocol (MCP) is an open standard, created by Anthropic, for AI applications to connect to tools and data. Previously, **[MCP servers only ran locally](https://support.claude.com/en/articles/10949351-getting-started-with-local-mcp-servers-on-claude-desktop)** (i.e. on a user's laptop). Now, developers can build and host remote MCP servers that communicate with AI apps over the internet. Remote MCP servers give models access to internet-hosted tools and data, transforming Claude into an informed teammate that can independently handle complex, multi-step projects tailored to your needs. ------------------------------------------------------------------------ ## Network requirements When you add a custom connector, Claude connects to your remote MCP server from Anthropic's cloud infrastructure, rather than from your local device. This is true across every Claude client, including claude.ai, Claude Desktop, Cowork, and the mobile apps. This means your MCP server must be reachable over the public internet from Anthropic's IP ranges. Servers hosted on a private corporate network, behind a VPN, or blocked by a firewall won't connect, even if you can reach them from your own machine. ### If your server is on a private network You’ll need to allowlist Anthropic's IP addresses in your firewall so inbound connections from Claude can reach your server. See **[Anthropic IP addresses](https://platform.claude.com/docs/en/api/ip-addresses)** for the current ranges. ### Why this applies to Cowork and Claude Desktop Even though Cowork and Claude Desktop run on your computer, remote connectors are configured and brokered through your Claude account. The connection to your MCP server originates from Anthropic's servers, not from your machine's network interface. Local MCP servers configured in Claude Desktop via `claude_desktop_config.json` are a separate mechanism and do use your local network, but those aren't available in Cowork or claude.ai. ------------------------------------------------------------------------ ## Add a custom connector **Note:** While anyone can build and host connectors using remote MCP, only Owners can add them to Team and Enterprise plans. Once a connector has been added to a Team or Enterprise organization, users individually connect to and enable that connector. This ensures that Claude can only access tools and data that the individual user has access to. ### For Team and Enterprise plans **Preliminary steps for owners:** Before members of Team and Enterprise plans can configure custom connectors, an Owner or Primary Owner needs to follow these initial steps to add a custom connector to your organization: 1. Navigate to **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. 2. Click the "Add" button. 3. Hover over “Custom,” then select “Web.” 4. Add your connector's remote MCP server URL. 5. Optionally, click “Advanced settings” to specify an OAuth Client ID and OAuth Client Secret for your server. 6. Finish configuring your connector by clicking "Add." **Steps for members after connector is configured:** 1. Navigate to **[Customize \> Connectors](https://claude.ai/customize/connectors)** 2. Find the custom connector your Owner added in the list (it will have a "Custom" label). 3. Click "Connect" to authenticate and start using the connector with Claude. ### For Pro and Max plans If you are using an individual Pro or Max plan, follow these steps to add a custom connector: 1. Navigate to **[Customize \> Connectors](https://claude.ai/customize/connectors)** 2. Click "+" then “Add custom connector.” 3. Add your connector's remote MCP server URL. 4. Optionally, click “Advanced settings” to specify an OAuth Client ID and OAuth Client Secret for your server. 5. Finish configuring your connector by clicking "Add." ### Enabling connectors after configuration You can enable connectors for individual conversations via the “+” button on the lower left of your chat interface, then "Connectors." You'll see your configured connectors with toggles allowing you to enable/disable them per conversation. ------------------------------------------------------------------------ ## Remove custom connectors You can remove a custom connector by following these steps: 1. Navigate to **[Customize \> Connectors](https://claude.ai/customize/connectors)** 1. Team and Enterprise Owners can do this on their organization's behalf in **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)** 2. Locate the "Connectors" section. 3. Click "Remove" or select the three dots next to the connector you'd like to remove. 4. Follow the prompts to remove. If you're hoping to edit a custom connector, you'll need to remove it first, then re-add it using the updated details. ------------------------------------------------------------------------ ## Build custom connectors To learn about building connectors to use with Claude, see **[Building custom connectors via remote MCP servers](https://support.claude.com/en/articles/11503834-building-custom-connectors-via-remote-mcp-servers)**. ------------------------------------------------------------------------ ## Security and privacy considerations Custom connectors allow you to connect Claude to arbitrary services that have not been verified by Anthropic. When you connect Claude to external services, you're granting it the ability to access and potentially modify data within those services based on your permissions. It’s important to make sure you’re only connecting to remote MCP servers that you trust and that you’re aware of Claude’s interactions with web connectors. ### Security and permissions When you add a custom connector to Claude, you'll typically go through an OAuth authentication process to securely sign in to the application and grant specific permissions. This allows Claude to interact with the application on your behalf, without Claude ever seeing your actual password. You can revoke these permissions at any time by disconnecting the connector in Claude's settings or the third-party service's security settings. Remote MCP servers act as intermediaries between Claude and external applications. You should: - **Only connect to trusted servers:** Only connect Claude to servers built and hosted by organizations and applications you trust. - **Review requested permissions carefully:** During auth, review what permissions the MCP server is requesting to the application. Limit these scopes when possible and deny access if requested permissions seem unnecessary. - **Be aware of prompt injections:** Malicious MCP servers may include hidden instructions that try to make Claude perform unintended actions. Claude has built-in protections that attempt to block these attacks, but it's important to pay attention to tool inputs & outputs and connect only to trusted servers. - **Monitor changes in tool behavior:** Server developers may update tool behavior unexpectedly, leading to unintended or malicious behavior. ### Reporting malicious MCP servers If you become aware of a malicious MCP server, please report it to our **[vulnerability disclosure program](https://hackerone.com/anthropic-vdp/)**, and choose [`https://github.com/modelcontextprotocol`](https://github.com/modelcontextprotocol) as the Asset. ### Taking actions with tools Remote MCP servers give Claude tools it can invoke during your conversation. The developer of an MCP server can define what these tools do, including: - Reading data from connected applications. - Creating, modifying, or deleting data in connected applications. - Taking actions on behalf of the user. Claude can only access resources that you've given the server permission to access, but you should: - Be aware of any actions Claude is taking and that they have no destructive or unintended effects. - Review Claude's tool approval requests carefully and only click "Allow always" when using a server and tool that you trust to run unsupervised. - Using the "Search and tools" menu, disable any tools that aren't relevant to the current conversation or that you don't want Claude to be able to invoke. ### Interactive connectors Some connectors can display interactive interfaces directly within your Claude conversations. Instead of only returning text-based responses, these connectors can open live, interactive apps — like dashboards, task boards, or design tools — right in the chat. Interactive connectors appear in two ways: - **Inline cards:** Compact components embedded in the conversation, showing summaries, confirmations, or quick actions. - **Fullscreen view:** Immersive interfaces for complex interactions like data visualizations or document editing. The conversation composer remains available so you can continue chatting with Claude. You can interact with these connectors directly — filtering data, checking off tasks, adjusting settings — without leaving the conversation. Any actions you take within the interface use the same permissions you granted when connecting the tool. **Admin controls:** Team and Enterprise plan owners can disable specific tool calls that render interactive connectors within **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. ### Using Claude with Research **Note: [Advanced Research](https://claude.com/blog/integrations)** is not currently able to invoke tools from local MCP servers. Research allows Claude to deeply investigate queries by searching through hundreds of internal and external sources. During the research process, Claude can invoke tools from your connectors automatically without further approval. When using Research with custom connectors: - Disable any tools that can take write actions in external applications. - Review Claude’s approval request carefully and be aware of which tools you’re granting Claude permission to invoke. - Be mindful of the impact of Claude sending a large number of requests to your connectors. See **[Using Research on Claude](https://support.claude.com/en/articles/11088861-using-research-on-claude)** for more information about this feature. --- ## about-integrations-using-remote-mcp.md # Get started with custom connectors using remote MCP Custom connectors using remote MCP are available on Claude and Claude Desktop for users on free, Pro, Max, Team, and Enterprise plans. Free users are limited to one custom connector. This feature is currently in beta. ## What are custom connectors? Custom connectors let you connect Claude directly to the tools and data sources that matter most to your workflows. This enables Claude to operate within your favorite software and draw insights from the complete context of your external tools. You can: - Connect Claude to existing remote MCP servers. - Build your own remote MCP servers to connect with any tool. **⚠️ Security and Privacy with Custom Connectors (beta)** Be aware that custom connectors allow you to connect Claude to services that have not been verified by Anthropic, and allow Claude to access and take action in these services. For more guidance, review the **[Security and privacy considerations](#h_9088ccdf4d)** section below. ## What are remote MCP servers? The Model Context Protocol (MCP) is an open standard, created by Anthropic, for AI applications to connect to tools and data. Previously, **[MCP servers only ran locally](https://support.claude.com/en/articles/10949351-getting-started-with-local-mcp-servers-on-claude-desktop)** (i.e. on a user's laptop). Now, developers can build and host remote MCP servers that communicate with AI apps over the internet. Remote MCP servers give models access to internet-hosted tools and data, transforming Claude into an informed teammate that can independently handle complex, multi-step projects tailored to your needs. ## Add a custom connector **Note:** While anyone can build and host connectors using remote MCP, only Owners can add them to Team and Enterprise plans. Once a connector has been added to a Team or Enterprise organization, users individually connect to and enable that connector. This ensures that Claude can only access tools and data that the individual user has access to. ### For Team and Enterprise plans **Preliminary steps for owners:** Before members of Team and Enterprise plans can configure custom connectors, an Owner or Primary Owner needs to follow these initial steps to add a custom connector to your organization: 1. Navigate to **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. 2. Click "Add custom connector" at the bottom of the section. 3. Add your connector's remote MCP server URL. 4. Optionally, click “Advanced settings” to specify an OAuth Client ID and OAuth Client Secret for your server. 5. Finish configuring your connector by clicking "Add." **Steps for members after connector is configured:** 1. Navigate to **[Settings \> Connectors](https://claude.ai/settings/connectors)**. 2. Locate the "Connectors" section. 3. Find the custom connector your Owner added in the list (it will have a "Custom" label). 4. Click "Connect" to authenticate and start using the connector with Claude. ### For Pro and Max plans If you are using an individual Pro or Max plan, follow these steps to add a custom connector: 1. Navigate to **[Settings \> Connectors](https://claude.ai/settings/connectors)**. 2. Locate the "Connectors" section. 3. Click "Add custom connector" at the bottom of the section. 4. Add your connector's remote MCP server URL. 5. Optionally, click “Advanced settings” to specify an OAuth Client ID and OAuth Client Secret for your server. 6. Finish configuring your connector by clicking "Add." ### Enabling connectors after configuration You can enable connectors for individual conversations via the “+” button on the lower left of your chat interface, then "Connectors." You'll see your configured connectors with toggles allowing you to enable/disable them per conversation. ## How to remove connectors You can remove a custom connector by following these steps: 1. Navigate to **[Settings \> Connectors](https://claude.ai/settings/connectors)** 1. Team and Enterprise Owners can do this on their organization's behalf in **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)** 2. Locate the "Connectors" section. 3. Click "Remove" or select the three dots next to the connector you'd like to remove. 4. Follow the prompts to remove. If you're hoping to edit a custom connector, you'll need to remove it first, then re-add it using the updated details. ## How to build custom connectors To learn about building connectors to use with Claude, see Building custom connectors via remote MCP servers. ## Security and privacy considerations Custom connectors allow you to connect Claude to arbitrary services that have not been verified by Anthropic. When you connect Claude to external services, you're granting it the ability to access and potentially modify data within those services based on your permissions. It’s important to make sure you’re only connecting to remote MCP servers that you trust and that you’re aware of Claude’s interactions with web connectors. ### Security and permissions When you add a custom connector to Claude, you'll typically go through an OAuth authentication process to securely sign in to the application and grant specific permissions. This allows Claude to interact with the application on your behalf, without Claude ever seeing your actual password. You can revoke these permissions at any time by disconnecting the connector in Claude's settings or the third-party service's security settings. Remote MCP servers act as intermediaries between Claude and external applications. You should: - **Only connect to trusted servers:** Only connect Claude to servers built and hosted by organizations and applications you trust. - **Review requested permissions carefully:** During auth, review what permissions the MCP server is requesting to the application. Limit these scopes when possible and deny access if requested permissions seem unnecessary. - **Be aware of prompt injections:** Malicious MCP servers may include hidden instructions that try to make Claude perform unintended actions. Claude has built-in protections that attempt to block these attacks, but it's important to pay attention to tool inputs & outputs and connect only to trusted servers. - **Monitor changes in tool behavior:** Server developers may update tool behavior unexpectedly, leading to unintended or malicious behavior. ### Reporting malicious MCP servers If you become aware of a malicious MCP server, please report it to our **[vulnerability disclosure program](https://hackerone.com/anthropic-vdp/)**, and choose [`https://github.com/modelcontextprotocol`](https://github.com/modelcontextprotocol) as the Asset. ### Taking actions with tools Remote MCP servers give Claude tools it can invoke during your conversation. The developer of an MCP server can define what these tools do, including: - Reading data from connected applications. - Creating, modifying, or deleting data in connected applications. - Taking actions on behalf of the user. Claude can only access resources that you've given the server permission to access, but you should: - Be aware of any actions Claude is taking and that they have no destructive or unintended effects. - Review Claude's tool approval requests carefully and only click "Allow always" when using a server and tool that you trust to run unsupervised. - Using the "Search and tools" menu, disable any tools that aren't relevant to the current conversation or that you don't want Claude to be able to invoke. ### Interactive connectors Some connectors can display interactive interfaces directly within your Claude conversations. Instead of only returning text-based responses, these connectors can open live, interactive apps — like dashboards, task boards, or design tools — right in the chat. Interactive connectors appear in two ways: - **Inline cards:** Compact components embedded in the conversation, showing summaries, confirmations, or quick actions. - **Fullscreen view:** Immersive interfaces for complex interactions like data visualizations or document editing. The conversation composer remains available so you can continue chatting with Claude. You can interact with these connectors directly — filtering data, checking off tasks, adjusting settings — without leaving the conversation. Any actions you take within the interface use the same permissions you granted when connecting the tool. **Admin controls:** Team and Enterprise plan owners can disable specific tool calls that render interactive connectors within **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. ### Using Claude with Research **Note:** **[Advanced Research](https://claude.com/blog/integrations)** is not currently able to invoke tools from local MCP servers. Research allows Claude to deeply investigate queries by searching through hundreds of internal and external sources. During the research process, Claude can invoke tools from your connectors automatically without further approval. When using Research with custom connectors: - Disable any tools that can take write actions in external applications. - Review Claude’s approval request carefully and be aware of which tools you’re granting Claude permission to invoke. - Be mindful of the impact of Claude sending a large number of requests to your connectors. See **[Using Research on Claude](../15-Claude-AI-Features/using-research-on-claude.md)** for more information about this feature. --- ## anthropic-connectors-directory-faq-e8c155aac1.md # Anthropic Connectors Directory FAQ May 7, 2026 The Connectors Directory FAQ has moved to the Claude developer docs. [Connectors overview →](https://claude.com/docs/connectors/overview) [Building custom connectors →](https://claude.com/docs/connectors/building) [Submitting to the Connectors Directory →](https://claude.com/docs/connectors/building/submission) --- ## anthropic-connectors-directory-faq.md # Anthropic Connectors Directory FAQ ## What is MCP? MCP stands for Model Context Protocol, an open standard created by Anthropic that allows AI applications to connect to tools and data sources. For comprehensive information about MCPs and how to get started, please refer to our remote MCP guide: [Getting Started with Custom Connectors Using Remote MCP](getting-started-with-custom-connectors-using-remote-mcp.md). ## What is the Connectors Directory? The [Connectors Directory](https://claude.com/connectors) aims to showcase Model Context Protocol servers that work with Claude across all our platforms - Claude web, Claude Desktop, our Claude mobile apps, Claude Code, and our API. The Connectors Directory serves as a single hub where users can discover MCP servers that Anthropic has reviewed. ### How do I get my MCP listed in the directory? You can fill out our [Connectors Directory server review form](https://docs.google.com/forms/d/e/1FAIpQLSeafJF2NDI7oYx1r8o0ycivCSVLNq92Mpc1FPxMKSw1CzDkqA/viewform) and we'll reach out about next steps if your MCP would be a good fit for the directory at this time. ### Are there standards my MCP will need to adhere to for inclusion in the directory? Yes, servers must adhere to the security, safety, and compatibility standards laid out in our [MCP Directory Policy](../22-Safety-Policy/anthropic-mcp-directory-policy.md). Adherence to these standards does not guarantee inclusion in the directory. ### I filled out the submission form but haven't heard back yet, how long will it take? While we strive to review every submission as quickly as we can, due to overwhelming interest we cannot promise that we will accept your submission or respond to it individually. ## 1. Authentication and Authorization ### Is OAuth 2.0 required if we don't need user authentication? OAuth 2.0 is only required if authentication is needed for the MCP server. Servers that do not require authentication do not require OAuth. ### Does Claude support machine-to-machine (client credentials) OAuth flow? No. Pure client credentials flow (machine-to-machine OAuth with just client_id/client_secret and no user interaction) is not supported. We can provision a static client_id/client_secret for Claude, but users must still complete an OAuth consent flow to authenticate their individual accounts. This works on claude.ai and Claude Desktop, but not Claude Code because Claude Code establishes connections directly from user’s computer to your MCP server (and thus don’t have access to the static credentials). ### Can we identify individual users without implementing OAuth? No. OAuth is the only way to uniquely identify users. We do not forward IP addresses, user IDs, or other metadata from end-users to MCP servers. ### Where does MCP server communication happen - client-side or Anthropic servers? It depends on the product surface. Communication occurs through Anthropic servers (claude.ai, Claude Desktop), user's browser, or user's local clients (Claude Code). There is no centralized connection method across all surfaces currently. ### Can users authenticate with personal credentials from within a corporate Claude account? Yes. Users authenticate with MCP servers on a per-server basis. They can use personal authentication credentials for MCP servers even when logged into claude.ai with a corporate account. ### Can we implement per-tool OAuth instead of per-server OAuth? No. OAuth occurs when the client connects to the MCP server, not at the tool level. You can enforce tool-level permissions based on user authentication after server connection is established. ### When a user disconnects an MCP tool, are OAuth tokens revoked? Tokens are removed from Anthropic's systems but not from your own systems. Access and refresh tokens at your identity provider remain valid until expiration, and session cookies are not cleared (as they are browser-level, outside Claude's control). ### How can our MCP service signal to Claude that a user's session is invalid? Return a 401 Unauthorized response. To trigger full regeneration of DCR client credentials, return an invalid_client error as per RFC 6749 section 5.2. ### Are client_id/client_secret required for standard OAuth flows? No. Client credentials are not required for standard OAuth flows with user authentication. ## 2. Rate Limiting and Abuse Prevention ### How can we implement per-user rate limiting without OAuth? Per-user rate limiting requires OAuth. Without OAuth, your options are: - Option 1: No authentication with global rate limiting - Option 2: IP allowlisting using Claude's static IPs ### Can we use IP allowlisting to restrict MCP server access to Claude users only? We strongly recommend using OAuth and not using IP allowlisting. However, this is possible using the static IP addresses available at [https://docs.claude.com/en/api/ip-addresses](https://docs.claude.com/en/api/ip-addresses). Important limitations: - Only works for claude.ai and Claude Desktop. Your MCP server will not work in Claude Code. - Egress IPs could change - advanced notice of change will be posted in docs - Not recommended as sole security measure, but okay to use when OAuth is not feasible ### Does Claude provide rate limiting or DDoS protection for MCP servers? No. MCP server owners must implement their own rate limiting and abuse prevention measures. ## 3. Technical Limits and Constraints ### What is the maximum token size for MCP tool results? 25,000 tokens. Use pagination, filtering, or limit parameters to reduce response size if needed. ### What are the timeout limits for MCP tools? Timeouts vary by component: - Claude Code: Configurable via MCP_TOOL_TIMEOUT setting - Claude.ai and Claude Desktop: 300 seconds (5 minutes) ## 4. Directory Submission and Requirements ### Which Claude plan types can access Connectors? All users can access connectors, except for custom connectors, which are available for paid plans only (Pro, Max, Team, and Enterprise). ### Do we need a specific Claude plan to submit an MCP to the directory? No. Any organization can submit to the directory regardless of plan type. ### Do we need to provide Anthropic with a test account? If your server requires authentication / an account to use, we will need you to provide us with a testing account for initial QA purposes. Including dummy data in the account is also helpful to check functionality. ### How many tools should an MCP server provide? No minimum or maximum requirement. We recommend that you start with a useful set of tools and expand capabilities over time. ### What are the performance requirements for MCP servers? There are currently no specific SLA requirements. However, connectors with performance or reliability issues may be removed from the directory. ### What are the most common reasons MCP submissions are delayed or rejected? Top issue: Missing tool annotations (30% of rejections) Requirement: All tools MUST include readOnlyHint or destructiveHint Impact: Immediate rejection, requires code changes. Not optional: Compliance requirement per MCP Directory Policy Example: { "name": "read_data", "readOnlyHint": true } ## 5. Data Privacy and Legal ### What telemetry does Anthropic collect from MCP tool calls? Telemetry includes all parameters and data passed into tool calls as well as the response from MCP server. ### Is MCP data used for training Claude models? We only train on consumer data when users explicitly opt in. See [https://www.anthropic.com/news/updates-to-our-consumer-terms](../19-Reference/updates-to-consumer-terms-and-privacy-policy-anthropic.md) for details. We do not train on Team/Enterprise plans. ### Can MCP servers identify which Claude SKU the user is on? No. MCP servers cannot determine the Claude SKU being used. ### What does 'Data Processing Agreement URL' mean in the submission form? Provide a URL to your company's data processing terms applicable to the MCP service. This does not need to be MCP-specific - your standard DPA is acceptable. ### Are connectors available globally? Can we restrict by geography? Connectors are available to Claude users globally. Geographic restriction is not supported. You can guide the user and Claude’s understanding of any geographic limitations to your MCP description or tool descriptions if needed (e.g., 'only U.S. organizations available'). ### Where can I find the MCP Directory Terms of Service and data policies? Key resources: - [MCP Directory Terms](../22-Safety-Policy/anthropic-mcp-directory-terms-and-conditions.md) - [MCP Directory Policy](../22-Safety-Policy/anthropic-mcp-directory-policy.md) ### What are data protection (e.g. GDPR, CCPA, LGPD) compliance requirements for MCP servers? MCP servers must handle compliance independently from Claude. If a user exercises a legal data protection right with your service, you must handle this on your end. ### Where must I include privacy policies for local connectors (desktop extensions)? In your .mcpb, there are two required locations: 1\. README.md - Add a "Privacy Policy" section with URL or full text. 2\. manifest.json - Add privacy_policies array with HTTPS URLs (requires manifest_version 0.2+) Example: { "manifest_version": "0.2", "privacy_policies": ["https://your-domain.com/privacy"], ... } **Must cover:** What data is collected, how it's used/stored, third-party sharing, retention, and contact info. ⚠️ Missing privacy policy will lead to an immediate rejection. ## 6. User Experience and Branding ### How do we update our connector's icon? We use the Google provided favicon, so you can customize your connector’s icon URL by updating the favicon for your connector URL. You can verify the icon that we will use at [https://www.google.com/s2/favicons?domain=\&sz=64](https://www.google.com/s2/favicons?domain=%3CYOUR_CONNECTOR_URL%3E&sz=64) Be sure to test each favicon size (16, 32, 48, 64, 96, 128) by updating the &sz= portion of the URL. **Remote connectors:** - If using a custom icon URL: Update the icon at your provided URL. - If relying on auto-detection: Icons are pulled from the Google-indexed favicon at your MCP server URL. Update your favicon and wait for Google to re-index. Otherwise, provide us with a URL with the favicon you wish to use as your icon. **Local connectors (desktop extensions):** Replace \`icon.png\` in your bundle and redistribute the updated \`.mcpb\` package to users. ### Does the connector descriptions in Claude support markdown? Yes. Markdown formatting is supported in connector descriptions. ### Can we include account requirements and sign-up links in the connector details? Yes. Please include account requirement language (e.g., 'free registered account required') and hyperlinks to sign-up pages in the connector details. ### How can we ensure users see our Terms of Service before connecting? Implement an OAuth screen. This is the only way to guarantee TOS display. Without OAuth, users could theoretically add connectors to Claude Desktop by directly editing claude_desktop_config.json. ### Can we show consent language with TOS/Privacy Notice links during installation? Include language in the connector description, but there is no special consent flow or popup. Use an OAuth screen to ensure users see and agree to terms. ## 7. Technical Implementation ### Is SSE transport acceptable, or must we use Streamable HTTP? You must use Streamable HTTP. ### Can we add server-level descriptions or only tool-level descriptions? Only tool-level descriptions are currently supported. No server-level description exists. This requires repeating steering guidance across individual tool descriptions. We recognize this is inconvenient and are exploring improvements. ### How do we make local MCP servers (.mcpb) available in Claude Code? Use the [plugin marketplace](../08-Plugins-Skills/create-and-distribute-a-plugin-marketplace-claude-code-docs-34a8f3c5e8.md). Node.js is recommended over Python or other languages for local MCPs due to better compatibility with Claude Desktop's built-in runtime. ### What are the observability best practices for local MCPs in Claude Desktop? Local MCPs can only log to ~/Library/Logs/Claude/mcp\*.log. Users must manually share log files for troubleshooting. Direct log transmission from local MCP servers is not supported due to security constraints. ## 8. Testing and Debugging ### IP allowlisting isn't working. How do we debug connection issues? Temporarily remove the IP allowlist to verify other configuration is correct. If connection works without allowlist, the issue is with IP filtering configuration. ### What testing should we complete before submitting to the directory? Essential testing prevents 80% of submission delays. See our guides for submitting remote and local connectors for best-practices and a pre-submission checklist. ## 9. Common Implementation Patterns ### What authentication approach should we use for our MCP? Choose based on your security requirements and constraints: ***Option 1: No Auth + Global Rate Limiting*** Best for: Public data with low abuse risk - Simplest implementation, works across all Claude surfaces - No per-user rate limiting, accessible by public internet - Can add IP allowlisting to restrict to Claude-only traffic ***Option 2: Full OAuth with User Registration*** Best for: Enterprise security, sensitive data, per-user rate limiting - Most secure, works across all surfaces, full user identification - Requires user registration system, more complex implementation - Recommended approach for production systems ***Option 3: IP-allowlisting*** We strongly recommend Option 2 over Option 3, but if implementing OAuth is truly infeasible, you can still get some restricted access to Claude users only - No ability to identify individual users, but restricts to requests from [claude.ai](http://claude.ai) and Claude Desktop. - MCP server will not work with Claude Code - Implementation: Use [Claude Static IPs](https://docs.claude.com/en/api/ip-addresses) ## 10. Key Resources Important links: - Claude Static IPs: [https://docs.claude.com/en/api/ip-addresses](https://docs.claude.com/en/api/ip-addresses) - MCP Directory Submission Forms: [https://docs.google.com/forms/d/e/1FAIpQLSeafJF2NDI7oYx1r8o0ycivCSVLNq92Mpc1FPxMKSw1CzDkqA/viewform](https://docs.google.com/forms/d/e/1FAIpQLSeafJF2NDI7oYx1r8o0ycivCSVLNq92Mpc1FPxMKSw1CzDkqA/viewform) - MCP Directory Terms: - MCP Directory Policy: - Training Opt-in Policy: [https://www.anthropic.com/news/updates-to-our-consumer-terms](../19-Reference/updates-to-consumer-terms-and-privacy-policy-anthropic.md) - MCP Protocol Specification: [https://modelcontextprotocol.io](https://modelcontextprotocol.io) - Local Desktop MCPB Bundling Utility: [https://github.com/anthropics/mcpb](https://github.com/anthropics/mcpb) - MCPB Manifest Specification: [https://github.com/anthropics/mcpb/blob/main/MANIFEST.md](https://github.com/anthropics/mcpb/blob/main/MANIFEST.md) - [Claude Desktop Release Notes](../19-Reference/release-notes.md) --- ## authorize-mcp-connectors-for-your-entire-organization-43f920f25a.md # Authorize MCP connectors for your entire organization June 18, 2026 This article explains how Enterprise-managed auth works and how admins can authorize connectors for their organization through their identity provider. With Enterprise-managed auth, you authorize a connector once for your entire organization, and your team inherits access automatically on first login. This feature is available in beta for Team and Enterprise plans on Claude. If you are a Claude customer, **[apply for access to get started](https://claude.com/form/ema-waitlist)**. If you are an MCP provider, **[apply here](https://docs.google.com/forms/d/e/1FAIpQLSf1goHGNDVFK7rncYuh6wnRpWSy7eGOcgL1i8uw3oyKFO9UUA/viewform?usp=sharing&ouid=101055591948883487705)**. We’ll share documentation with customers and MCP providers on how to get started once you have access. ## What is Enterprise-managed auth? Enterprise-managed auth is an authorization and authentication model for connectors in Claude. Instead of having every person authenticate each connector individually, admins provision connector access centrally through the organization's identity provider. Once you enable a connector for your organization, your team gets it automatically the first time they log in, with identity inherited from their existing identity provider groups and roles. ## What you control You decide which connectors are enabled, which groups or roles get them, and at what access level: - Auth connectors once for your organization, and access is given to your team automatically. - Scope access by group, team, or role, so different teams get the access that fits their work (integration with role-based permissions coming soon). - Revoke access by deprovisioning someone in your identity provider, which removes their connector access at the same time. - Require that a connector only ever connects through your identity provider so personal accounts stay out of work tools. **Note:** Your identity provider and each connector are operated by third parties under their own terms. Claude relays the authorization your identity provider issues; access decisions, scoping, and the data each connector can reach are governed by your identity provider’s policies and the connected service’s permissions, not by Anthropic. Token lifetimes and lifecycle are managed by the connected authorization server and identity provider. Existing sessions end when the connector’s access token expires or is revoked. ## What works with Enterprise-managed auth Enterprise-managed auth brings together two things your organization already uses: your identity provider, which controls who gets access, and the connectors your teams work with day to day. ### Identity providers Okta is supported at launch, with more identity providers coming soon. See **[Okta’s documentation](https://support.okta.com/help/s/article/claude-enterprise-managed-auth-with-okta-cross-app-access-xaa-beta-participation-guide?language=en_US)** for more details. ### Connectors Currently, you can provision these connectors through Enterprise-managed auth: - Asana - Atlassian - Canva - Figma - Granola - Linear - Supabase - Slack (coming soon) Any MCP provider can add support for Enterprise-managed auth. See **[Enterprise-Managed Authorization](https://modelcontextprotocol.io/extensions/auth/enterprise-managed-authorization)** for more details. ### Personal connectors Your team can still add personal connectors on top of what you provision. Enterprise-managed auth handles the connectors you enable for your organization, while individuals can connect additional services for their own use. --- ## browse-skills-connectors-and-plugins-in-one-directory-6ee3c0df4b.md # Browse skills, connectors, and plugins in one directory May 29, 2026 Our unified directory brings skills, connectors, and plugins together in one place so you can find and install everything that customizes Claude without switching between separate menus. On Team and Enterprise plans, the directory is also where skills shared across your organization appear. ------------------------------------------------------------------------ ## Access the directory 1. Open Claude or Claude Desktop 2. Click “Customize” in the left sidebar. 3. Click on the tab you want to view (Skills, Connectors, or Plugins), then the “+” button. 4. Select “Browse \[skills / connectors / plugins\]” to open the unified directory. ### Skills To open the Skills directory, go to **[Customize \> Skills](https://claude.ai/customize/skills)**, click the “+” button, then “Browse skills.” Click the “Install” button on any skill to add it to your skills list. Once installed, the skill appears in **[Customize \> Skills](https://claude.ai/customize/skills)** and is enabled by default, so Claude can use it automatically when relevant. You can toggle it off at any time, but the skill stays in the directory so you can re-enable it later. Skills you install from the directory are view-only. You can use them, but you can't edit their contents. If you want to change how a skill works, download a copy, modify it, and upload it as your own. For members of Team and Enterprise plans, the “Your organization” tab surfaces skills that have been shared with your entire organization. This includes skills that owners have provisioned centrally and skills that colleagues have shared organization-wide. You can’t fully delete organization-shared skills from your list. Only the person who shared the skill or an organization owner can remove it from the directory. **Note:** Skills shared with you directly by a colleague don’t appear in the directory. They go straight to your skills list. See **[Use skills in Claude](https://support.claude.com/en/articles/12512180-use-skills-in-claude)** for details on how shared skills work. ### Connectors The connectors tab shows the same catalog of services you can connect Claude to, now in the unified view. Click "Connect" on any service to start the authentication flow. For more on connecting and managing services, see **[Use connectors to extend Claude’s capabilities](https://support.claude.com/en/articles/11176164-use-connectors-to-extend-claude-s-capabilities)**. ### Plugins The plugins tab shows available plugins, including any your organization has distributed through a marketplace. Click "Install" to add a plugin. Once installed, the skills from that plugin are available in chat (on the web and the Chat tab in Claude Desktop) and in Cowork. For more on installing and customizing plugins, see **[Use plugins in Claude](https://support.claude.com/en/articles/13837440-)**. ## For organization owners Skill sharing to the directory is off by default. To let people in your organization share skills org-wide, enable the **Share with organization** toggle in **[Organization settings \> Skills](https://claude.ai/admin-settings/skills)**. You can also enable the **Skill sharing** toggle separately, which lets people share skills with specific colleagues without publishing to the directory. For full details on configuring sharing and provisioning skills centrally, see **[Provision and manage skills for your organization](https://support.claude.com/en/articles/13119606-provision-and-manage-skills-for-your-organization)**. --- ## build-custom-connectors-via-remote-mcp-servers-a233674fe1.md # Build custom connectors via remote MCP servers April 15, 2026 The guide to building custom connectors has moved to the Claude developer docs. [View the building guide →](https://claude.com/docs/connectors/building) --- ## building-custom-connectors-via-remote-mcp-servers-813ea92951.md # Build custom connectors via remote MCP servers April 15, 2026 The guide to building custom connectors has moved to the Claude developer docs. [View the building guide →](https://claude.com/docs/connectors/building) --- ## building-custom-connectors-via-remote-mcp-servers.md 3. Building custom connectors via remote MCP servers # Building custom connectors via remote MCP servers Updated today Custom connectors using remote MCP are available on Claude, Cowork, and Claude Desktop for users on free, Pro, Max, Team, and Enterprise plans. Free users are limited to one custom connector. This feature is currently in beta. ## Building remote MCP servers To get started with remote servers, start with the following resources: - The [auth spec](../06-MCP-Tools/Spec-Archive/spec-2025-06-18-authorization-model-context-protocol.md), especially details on [the auth flow for third-party services](../06-MCP-Tools/Spec-Archive/spec-2025-03-26-authorization-model-context-protocol.md#2-10-third-party-authorization-flow). - The remote server examples in the [TypeScript](https://github.com/modelcontextprotocol/typescript-sdk/tree/main/src/examples/server) and [Python](https://github.com/modelcontextprotocol/python-sdk/tree/main/examples/servers) SDKs. - The client and server auth implementations in the [TypeScript](https://github.com/modelcontextprotocol/typescript-sdk/tree/main/src/server/auth) and [Python](https://github.com/modelcontextprotocol/python-sdk/tree/main/src/mcp) SDKs. - The official MCP [roadmap](../06-MCP-Tools/Community/roadmap-model-context-protocol.md) and [draft spec’s changelog](../06-MCP-Tools/Spec/spec-draft-key-changes-model-context-protocol.md) for details on how the protocol will evolve. Other resources (like [this](https://simplescraper.io/blog/how-to-mcp)) may also be helpful to learn about considerations when building, deploying, and troubleshooting remote servers. In addition, some [solutions like Cloudflare](https://developers.cloudflare.com/agents/guides/remote-mcp-server/) provide remote MCP server hosting with built-in autoscaling, OAuth token management, and deployment. ## MCP support ### Platforms - Remote MCP servers are supported on Claude and Claude Desktop for Pro, Max, Team, and Enterprise plans. - To configure remote MCP servers for use in Claude Desktop, add them via [Settings \> Connectors](https://claude.ai/settings/connectors). Claude Desktop will not connect to remote servers that are configured directly via *claude_desktop_config.json*. - As of July, Claude for iOS and Android also support remote MCP servers! - Users can use tools, prompts, and resources from remote servers that they’ve already added via [claude.ai](http://claude.ai). Users cannot add new servers directly from Claude Mobile. ### Transport and auth - Claude supports both SSE- and Streamable HTTP-based remote servers, although support for SSE may be deprecated in the coming months. - Claude supports both authless and OAuth-based remote servers. **Auth support** - Claude supports the [3/26 auth spec](../06-MCP-Tools/Spec-Archive/spec-2025-03-26-authorization-model-context-protocol.md#1-introduction) and (as of July) the [6/18 auth spec](../06-MCP-Tools/Spec-Archive/spec-2025-06-18-authorization-model-context-protocol.md). - Claude supports [Dynamic Client Registration](../06-MCP-Tools/Spec-Archive/spec-2025-03-26-authorization-model-context-protocol.md#2-4-dynamic-client-registration) (DCR). - OAuth servers can signal to Claude that a DCR client has been deleted and that Claude should re-register the client by returning an HTTP 401 with an error of invalid_client from the token endpoint, as described in [RFC 6749](https://datatracker.ietf.org/doc/html/rfc6749#section-5.2). - As of July, users are also able to specify a custom client ID and client secret when configuring a server that doesn’t support DCR. - Claude’s OAuth callback URL is [https://claude.ai/api/mcp/auth_callback](https://claude.ai/api/mcp/auth_callback) and its OAuth client name is Claude. - This callback URL may change to [https://claude.com/api/mcp/auth_callback](https://claude.com/api/mcp/auth_callback) in the future – if you choose to allowlist MCP client callback URLs, please allowlist this callback URL as well to ensure that your server continues to work with Claude. - Claude supports token expiry and refresh – servers should support this functionality in order to provide the best experience for users. See [here](https://docs.anthropic.com/en/api/ip-addresses#ipv4-2) for the IP addresses used by Claude for inbound and outbound connections to MCP servers. Server developers wishing to disallow non-Claude MCP Clients can whitelist these IP addresses, Claude’s OAuth callback URL, and/or Claude’s OAuth client name. ### Protocol Features - Claude supports tools, prompts, and resources. - Claude supports text- and image-based tool results. - Claude supports text- and binary- based resources. - Claude does not yet support resource subscriptions, sampling, and other more advanced or draft capabilities. ## Testing remote MCP servers The best way to test and validate a server is to try [adding it to Claude](https://support.anthropic.com/en/articles/11175166-getting-started-with-custom-connectors-using-remote-mcp#h_3d1a65aded). Alternatively, use the [inspector tool](https://github.com/modelcontextprotocol/inspector). This will allow you to validate: - that your server successfully initiates and completes the auth flow. - that your server correctly implements various parts of the auth flow. - which tools, prompts, resources, and other MCP features your server exposes. See the [MCP documentation](../06-MCP-Tools/Tutorials/mcp-inspector-model-context-protocol.md) for more details on using inspector and for other tips on how to debug and troubleshoot your server. In addition, other solutions like [Cloudflare’s AI Playground](https://playground.ai.cloudflare.com/) allow you to test remote MCP server functionality. --- ## claude-code-in-slack-claude-code-docs-c62a093e92.md # Claude Code in Slack Delegate coding tasks directly from your Slack workspace > ## Documentation Index > > Fetch the complete documentation index at: > > Use this file to discover all available pages before exploring further. Claude Code in Slack brings the power of Claude Code directly into your Slack workspace. When you mention `@Claude` with a coding task, Claude automatically detects the intent and creates a Claude Code session on the web, allowing you to delegate development work without leaving your team conversations. This integration is built on the existing Claude for Slack app but adds intelligent routing to Claude Code on the web for coding-related requests. Use cases - **Bug investigation and fixes**: Ask Claude to investigate and fix bugs as soon as they’re reported in Slack channels. - **Quick code reviews and modifications**: Have Claude implement small features or refactor code based on team feedback. - **Collaborative debugging**: When team discussions provide crucial context (e.g., error reproductions or user reports), Claude can use that information to inform its debugging approach. - **Parallel task execution**: Kick off coding tasks in Slack while you continue other work, receiving notifications when complete. Prerequisites Before using Claude Code in Slack, ensure you have the following: | Requirement | Details | |:-----------------------|:--------------------------------------------------------------------------------------------------| | Claude Plan | Pro, Max, Team, or Enterprise with Claude Code access (premium seats or Chat + Claude Code seats) | | Claude Code on the web | Access to [Claude Code on the web](/docs/en/claude-code-on-the-web) must be enabled | | GitHub Account | Connected to Claude Code on the web with at least one repository authenticated | | Slack Authentication | Your Slack account linked to your Claude account via the Claude app | Setting up Claude Code in Slack 1 Install the Claude App in Slack A workspace administrator must install the Claude app from the Slack App Marketplace. Visit the [Slack App Marketplace](https://slack.com/marketplace/A08SF47R6P4) and click “Add to Slack” to begin the installation process. 2 Connect your Claude account After the app is installed, authenticate your individual Claude account: 1. Open the Claude app in Slack by clicking on “Claude” in your Apps section 2. Navigate to the App Home tab 3. Click “Connect” to link your Slack account with your Claude account 4. Complete the authentication flow in your browser 3 Configure Claude Code on the web Ensure your Claude Code on the web is properly configured: - Visit [claude.ai/code](https://claude.ai/code) and sign in with the same account you connected to Slack - Connect your GitHub account if not already connected - Authenticate at least one repository that you want Claude to work with 4 Choose your routing mode After connecting your accounts, configure how Claude handles your messages in Slack. Navigate to the Claude App Home in Slack to find the **Routing Mode** setting. | Mode | Behavior | |:----------------|:-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------| | **Code only** | Claude routes all @mentions to Claude Code sessions. Best for teams using Claude in Slack exclusively for development tasks. | | **Code + Chat** | Claude analyzes each message and intelligently routes between Claude Code (for coding tasks) and Claude Chat (for writing, analysis, and general questions). Best for teams who want a single @Claude entry point for all types of work. | In Code + Chat mode, if Claude routes a message to Chat but you wanted a coding session, you can click “Retry as Code” to create a Claude Code session instead. Similarly, if it’s routed to Code but you wanted a Chat session, you can choose that option in that thread. 5 Add Claude to channels Claude is not automatically added to any channels after installation. To use Claude in a channel, invite it by typing `/invite @Claude` in that channel. Claude can only respond to @mentions in channels where it has been added. How it works Automatic detection When you mention @Claude in a Slack channel or thread, Claude automatically analyzes your message to determine if it’s a coding task. If Claude detects coding intent, it will route your request to Claude Code on the web instead of responding as a regular chat assistant. You can also explicitly tell Claude to handle a request as a coding task, even if it doesn’t automatically detect it. Claude Code in Slack only works in channels (public or private). It does not work in direct messages (DMs). Context gathering **From threads**: When you @mention Claude in a thread, it gathers context from all messages in that thread to understand the full conversation. **From channels**: When mentioned directly in a channel, Claude looks at recent channel messages for relevant context. This context helps Claude understand the problem, select the appropriate repository, and inform its approach to the task. When @Claude is invoked in Slack, Claude is given access to the conversation context to better understand your request. Claude may follow directions from other messages in the context, so users should make sure to only use Claude in trusted Slack conversations. Session flow 1. **Initiation**: You @mention Claude with a coding request 2. **Detection**: Claude analyzes your message and detects coding intent 3. **Session creation**: A new Claude Code session is created on claude.ai/code 4. **Progress updates**: Claude posts status updates to your Slack thread as work progresses 5. **Completion**: When finished, Claude @mentions you with a summary and action buttons 6. **Review**: Click “View Session” to see the full transcript, or “Create PR” to open a pull request User interface elements App Home The App Home tab shows your connection status and allows you to connect or disconnect your Claude account from Slack. Message actions - **View Session**: Opens the full Claude Code session in your browser where you can see all work performed, continue the session, or make additional requests. - **Create PR**: Creates a pull request directly from the session’s changes. - **Retry as Code**: If Claude initially responds as a chat assistant but you wanted a coding session, click this button to retry the request as a Claude Code task. - **Change Repo**: Allows you to select a different repository if Claude chose incorrectly. Repository selection Claude automatically selects a repository based on context from your Slack conversation. If multiple repositories could apply, Claude may display a dropdown allowing you to choose the correct one. Access and permissions User-level access | Access Type | Requirement | |:---------------------|:----------------------------------------------------------------| | Claude Code Sessions | Each user runs sessions under their own Claude account | | Usage & Rate Limits | Sessions count against the individual user’s plan limits | | Repository Access | Users can only access repositories they’ve personally connected | | Session History | Sessions appear in your Claude Code history on claude.ai/code | Workspace-level access Slack workspace administrators control whether the Claude app is available in their workspace: | Control | Description | |:-----------------------------|:------------------------------------------------------------------------------------------------------------------| | App installation | Workspace admins decide whether to install the Claude app from the Slack App Marketplace | | Enterprise Grid distribution | For Enterprise Grid organizations, organization admins can control which workspaces have access to the Claude app | | App removal | Removing the app from a workspace immediately revokes access for all users in that workspace | Channel-based access control Claude is not automatically added to any channels after installation. Users must explicitly invite Claude to channels where they want to use it: - **Invite required**: Type `/invite @Claude` in any channel to add Claude to that channel - **Channel membership controls access**: Claude can only respond to @mentions in channels where it has been added - **Access gating through channels**: Admins can control who uses Claude Code by managing which channels Claude is invited to and who has access to those channels - **Private channel support**: Claude works in both public and private channels, giving teams flexibility in controlling visibility This channel-based model allows teams to restrict Claude Code usage to specific channels, providing an additional layer of access control beyond workspace-level permissions. What’s accessible where **In Slack**: You’ll see status updates, completion summaries, and action buttons. The full transcript is preserved and always accessible. **On the web**: The complete Claude Code session with full conversation history, all code changes, file operations, and the ability to continue the session or create pull requests. For Enterprise and Team accounts, sessions created from Claude in Slack are automatically visible to the organization. See [Claude Code on the Web sharing](/docs/en/claude-code-on-the-web#share-sessions) for more details. Best practices Writing effective requests - **Be specific**: Include file names, function names, or error messages when relevant. - **Provide context**: Mention the repository or project if it’s not clear from the conversation. - **Define success**: Explain what “done” looks like—should Claude write tests? Update documentation? Create a PR? - **Use threads**: Reply in threads when discussing bugs or features so Claude can gather the full context. When to use Slack vs. web **Use Slack when**: Context already exists in a Slack discussion, you want to kick off a task asynchronously, or you’re collaborating with teammates who need visibility. **Use the web directly when**: You need to upload files, want real-time interaction during development, or are working on longer, more complex tasks. Troubleshooting Sessions not starting 1. Verify your Claude account is connected in the Claude App Home 2. Check that you have Claude Code on the web access enabled 3. Ensure you have at least one GitHub repository connected to Claude Code Repository not showing 1. Connect the repository in Claude Code on the web at [claude.ai/code](https://claude.ai/code) 2. Verify your GitHub permissions for that repository 3. Try disconnecting and reconnecting your GitHub account Wrong repository selected 1. Click the “Change Repo” button to select a different repository 2. Include the repository name in your request for more accurate selection Authentication errors 1. Disconnect and reconnect your Claude account in the App Home 2. Ensure you’re signed into the correct Claude account in your browser 3. Check that your Claude plan includes Claude Code access Session expiration 1. Sessions remain accessible in your Claude Code history on the web 2. You can continue or reference past sessions from [claude.ai/code](https://claude.ai/code) Current limitations - **GitHub only**: Currently supports repositories on GitHub. - **One PR at a time**: Each session can create one pull request. - **Rate limits apply**: Sessions use your individual Claude plan’s rate limits. - **Web access required**: Users must have Claude Code on the web access; those without it will only get standard Claude chat responses. Related resources ## Claude Code on the web Learn more about Claude Code on the web ## Claude for Slack General Claude for Slack documentation ## Slack App Marketplace Install the Claude app from the Slack Marketplace ## Claude Help Center Get additional support --- ## claude-for-microsoft-365-claude-by-anthropic-06f94a660e.md # Claude for Microsoft 365 Work without the window shuffle Claude works inside Excel, PowerPoint, Word, and Outlook. Start in your inbox, end in the deck. It remembers everything in between. Install for Microsoft 365 [Install for Microsoft 365](https://marketplace.microsoft.com/en-us/product/office/WA200010725?tab=Overview) Install for Microsoft 365 Install for Outlook [Install for Outlook](https://marketplace.microsoft.com/en-us/product/office/WA200010724?tab=Overview) Install for Outlook ## Claude for Excel Ask about any cell, update assumptions without breaking formulas, and build models from scratch. ## Claude for PowerPoint Build slides in your template, edit what you've selected, and generate native charts and diagrams. ## Claude for Word Edit with tracked changes, respond to comment threads, and update content using company styles. Beta ## Claude for Outlook Triage your inbox in one prompt, draft replies that wait for you to send, and find time across calendars. ## How teams use Claude for Microsoft 365 [Play video](#) Play video ### Turn an email into a deliverable Triage email, open the brief in Word, build the model in Excel, make the PowerPoint deck, draft the review invite. Context carries across each step. ### Clear your inbox before the meeting Ask Claude what needs your attention, get replies drafted for a few quick wins, and open the brief from your inbox. The context comes with it into Word. ### Build a model from a brief Ask Claude to build the model behind your Word brief. It opens Excel, pulls the numbers from the doc, and builds real formulas across multiple tabs. ### Draft in your template Point Claude at a brief, a dataset, or a thread. It drafts sections in your heading styles, populates tables in place, and keeps your numbering intact. ## The Claude you trust, inside Microsoft 365 ### Every edit is reviewable Tracked changes in Word, highlighted cells in Excel, drafts that wait in Outlook. Nothing goes out or gets saved until you say so. ### Your templates and formatting Claude uses your rules, whether for slide masters, heading styles, formula structures, or numbering so you stay on brand. ### Enterprise security, built in Sign in with a Claude account, or connect through your existing cloud provider. Works within your existing compliance framework. How can Claude help? Open the brief in Word and start the memo from our firm template. Copy prompt [Copy prompt](https://claude.ai/new?q=Open+the+brief+in+Word+and+start+the+memo+from+our+firm+template.) Copy prompt How can Claude help? Turn this updated analysis into the SteerCo deck in the Acme template. Copy prompt [Copy prompt](https://claude.ai/new?q=Turn+this+updated+analysis+into+the+SteerCo+deck+in+the+Acme+template) Copy prompt How can Claude help? Find 30 minutes with everyone on this thread next week Copy prompt [Copy prompt](https://claude.ai/new?q=What+did+the+other+side+change+in+section+4%2C+and+are+any+of+these+dealbreakers%3F) Copy prompt [Prev](#) Prev Blog post Claude for Excel, PowerPoint, and Word are now generally available on all paid plans, with Claude for Outlook joining in beta. Read blog [Read blog](/blog/collaborate-with-claude-across-excel-powerpoint-word-and-outlook) Read blog ### What customers are saying “Claude enables a seamless transition from rough ideas to polished, branded deliverables across presentation, document and spreadsheet applications without disrupting the workflow. That end-to-end consistency has been a significant time-saver, especially when quality, formatting, and speed all matter.” Vivek Kulkarni, US AI Transformation Leader “Claude in Excel allows my teams to build an initial version of complex models faster, enabling them to focus on refining the model, pressure testing inputs and assumption, and exploring more scenarios and trade-offs. This helps us have richer and deeper discussions with our clients earlier.” Gene Rapoport, Head of Private Equity AI Practice “Claude for M365 is being rapidly adopted by ServiceNow. Claude does the work in Excel itself, instead of asking us to move content between tools, driving step change in productivity.” Rajeev Sethi, GVP Enterprise Technologies “We can instruct Claude to build a style guide from an executive's prior written communications so our EAs can draft their emails in their voice. That's high-leverage work the team would never have time to set up by hand.” Ben Letalik, Sr. Director, Digital Transformation & Innovation “Our investment professionals live in data and analytical models, and Claude for Excel meets them there. Analysts are using it to build and update coverage models, separate signal from noise, and pressure-test their work — all with a step-change in efficiency.” Atte Lahtiranta, Head of Core Engineering [Prev](#) Prev 0/5 ## Core capabilities across the apps you use every day ### One conversation across all your apps Context carries from Outlook to Word to Excel to PowerPoint. Change one thing, and every open file catches up. ### Deploy through your preferred cloud Connect through Microsoft Foundry, Amazon Bedrock, or Google Cloud Vertex AI credentials. ### Bring any document into the conversation Drop a PDF or doc into any sidebar. Claude reads it alongside your open file and references it in other apps. ### Turn workflows into Skills When a process is right, save it as a skill. The team can use it the same way in all four apps. Learn more [Learn more](/skills) Learn more ### Type or talk, whichever is faster Use dictation to talk to Claude in a sidebar instead of typing. ### Access connectors Pull context from outside sources directly from the sidebar. Learn more [Learn more](/connectors) Learn more ## FAQ ### Where is this available? Claude for Excel, PowerPoint, and Word are generally available on all paid Claude plans. Claude for Outlook is in beta on all paid plans. ### Can I connect through my cloud provider? Yes. Claude for Microsoft 365 is available on Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry. ### Does Claude preserve my formatting and templates? Yes. Claude inherits your heading styles, slide masters, formula conventions, and numbering. Edit one section without disturbing the rest. ### Can Claude send an email without me seeing it first? No. Drafts and calendar invites open in Outlook's native compose form and wait for you to click send. [Prev](#) Prev ## Get started with Claude for Microsoft 365 Install for Microsoft 365 [Install for Microsoft 365](https://marketplace.microsoft.com/en-us/product/office/WA200010725?tab=Overview) Install for Microsoft 365 Install for Outlook [Install for Outlook](https://marketplace.microsoft.com/en-us/product/office/WA200010724?tab=Overview) Install for Outlook [Homepage](https://claude.com) Homepage Thank you! Your submission has been received! Oops! Something went wrong while submitting the form. Write [Button Text](#) Button Text Learn [Button Text](#) Button Text Code [Button Text](#) Button Text Write - Help me develop a unique voice for an audience Hi Claude! Could you help me develop a unique voice for an audience? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Improve my writing style Hi Claude! Could you improve my writing style? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Brainstorm creative ideas Hi Claude! Could you brainstorm creative ideas? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! Learn - Explain a complex topic simply Hi Claude! Could you explain a complex topic simply? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Help me make sense of these ideas Hi Claude! Could you help me make sense of these ideas? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Prepare for an exam or interview Hi Claude! Could you prepare for an exam or interview? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! Code - Explain a programming concept Hi Claude! Could you explain a programming concept? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Look over my code and give me tips Hi Claude! Could you look over my code and give me tips? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Vibe code with me Hi Claude! Could you vibe code with me? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! More - Write case studies This is another test - Write grant proposals Hi Claude! Could you write grant proposals? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to — like Google Drive, web search, etc. — if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can - an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Write video scripts this is a test [Anthropic](https://www.anthropic.com/) Anthropic © \[year\] Anthropic PBC Products - Claude [Claude](/product/overview) Claude - Claude Code [Claude Code](/product/claude-code) Claude Code - Claude Code for Enterprise [Claude Code for Enterprise](/product/claude-code/enterprise) Claude Code for Enterprise - Claude Cowork [Claude Cowork](/product/cowork) Claude Cowork - @Claude [@Claude](/product/tag) @Claude - Claude Design [Claude Design](/product/design) Claude Design - Claude Science [Claude Science](/product/claude-science) Claude Science - Claude Security [Claude Security](/product/claude-security) Claude Security - Download app [Download app](/download) Download app - Pricing [Pricing](/pricing) Pricing - Log in [Log in](https://claude.ai/login) Features - Claude for Chrome [Claude for Chrome](/claude-for-chrome) Claude for Chrome - Claude for Microsoft 365 [Claude for Microsoft 365](/claude-for-microsoft-365) Claude for Microsoft 365 - Skills [Skills](/skills) Skills Models - Mythos [Mythos](https://www.anthropic.com/claude/mythos) Mythos - Fable [Fable](https://www.anthropic.com/claude/fable) Fable - Opus [Opus](https://www.anthropic.com/claude/opus) Opus - Sonnet [Sonnet](https://www.anthropic.com/claude/sonnet) Sonnet - Haiku [Haiku](https://www.anthropic.com/claude/haiku) Haiku Solutions - AI agents [AI agents](/solutions/agents) AI agents - Code modernization [Code modernization](/solutions/code-modernization) Code modernization - Coding [Coding](/solutions/coding) Coding - Customer support [Customer support](/solutions/customer-support) Customer support - Cybersecurity [Cybersecurity](/solutions/cybersecurity) Cybersecurity - Enterprise [Enterprise](/solutions/enterprise) Enterprise - Financial services [Financial services](/solutions/financial-services) Financial services - Government [Government](/solutions/government) Government - Healthcare [Healthcare](/solutions/healthcare) Healthcare - Higher education [Higher education](/solutions/education) Higher education - K-12 teachers [K-12 teachers](/solutions/teachers) K-12 teachers - Legal [Legal](/solutions/legal) Legal - Life sciences [Life sciences](/solutions/life-sciences) Life sciences - Nonprofits [Nonprofits](/solutions/nonprofits) Nonprofits - Small business [Small business](/solutions/small-business) Small business Claude Platform - Overview [Overview](/platform/api) Overview - Developer docs [Developer docs](https://platform.claude.com/docs) Developer docs - Pricing [Pricing](https://claude.com/pricing#api) Pricing - Ecosystem [Ecosystem](/ecosystem) Ecosystem - Marketplace [Marketplace](/platform/marketplace) Marketplace - Claude on AWS [Claude on AWS](/partners/claude-on-aws) Claude on AWS - Google Cloud [Google Cloud](/partners/google-cloud) Google Cloud - Microsoft Foundry [Microsoft Foundry](/partners/microsoft-foundry) Microsoft Foundry - Regional compliance [Regional compliance](/regional-compliance) Regional compliance - Console login [Console login](https://platform.claude.com/) Console login Resources - Blog [Blog](/blog) Blog - Claude partner network [Claude partner network](/partners) Claude partner network - Community [Community](/community) Community - Connectors [Connectors](/connectors) Connectors - Courses [Courses](https://www.anthropic.com/learn) Courses - Customer stories [Customer stories](/customers) Customer stories - Engineering at Anthropic [Engineering at Anthropic](https://www.anthropic.com/engineering) Engineering at Anthropic - Events [Events](https://www.anthropic.com/events) Events - Plugins [Plugins](/plugins) Plugins - Powered by Claude [Powered by Claude](/partners/powered-by-claude) Powered by Claude - Service partners [Service partners](/partners/services) Service partners - Tutorials [Tutorials](/resources/tutorials) Tutorials - Use cases [Use cases](/resources/use-cases) Use cases Company - Anthropic [Anthropic](https://www.anthropic.com/) Anthropic - Careers [Careers](https://www.anthropic.com/careers) Careers - Policy [Policy](https://www.anthropic.com/policy) Policy - Economic Futures [Economic Futures](https://www.anthropic.com/economic-futures) Economic Futures - Research [Research](https://www.anthropic.com/research) Research - News [News](https://www.anthropic.com/news) News - Policy on the AI Exponential --- ## claude-for-microsoft-365-claude-by-anthropic-99f4da04a3.md # Claude for Microsoft 365 Work without the window shuffle Claude works inside Excel, PowerPoint, Word, and Outlook. Start in your inbox, end in the deck. It remembers everything in between. Install for Microsoft 365 [Install for Microsoft 365](https://marketplace.microsoft.com/en-us/product/office/WA200010725?tab=Overview) Install for Microsoft 365 Install for Outlook [Install for Outlook](https://marketplace.microsoft.com/en-us/product/office/WA200010724?tab=Overview) Install for Outlook ## Claude for Excel Ask about any cell, update assumptions without breaking formulas, and build models from scratch. ## Claude for PowerPoint Build slides in your template, edit what you've selected, and generate native charts and diagrams. ## Claude for Word Edit with tracked changes, respond to comment threads, and update content using company styles. Beta ## Claude for Outlook Triage your inbox in one prompt, draft replies that wait for you to send, and find time across calendars. ## How teams use Claude for Microsoft 365 [Play video](#) Play video ### Turn an email into a deliverable Triage email, open the brief in Word, build the model in Excel, make the PowerPoint deck, draft the review invite. Context carries across each step. ### Clear your inbox before the meeting Ask Claude what needs your attention, get replies drafted for a few quick wins, and open the brief from your inbox. The context comes with it into Word. ### Build a model from a brief Ask Claude to build the model behind your Word brief. It opens Excel, pulls the numbers from the doc, and builds real formulas across multiple tabs. ### Draft in your template Point Claude at a brief, a dataset, or a thread. It drafts sections in your heading styles, populates tables in place, and keeps your numbering intact. ## The Claude you trust, inside Microsoft 365 ### Every edit is reviewable Tracked changes in Word, highlighted cells in Excel, drafts that wait in Outlook. Nothing goes out or gets saved until you say so. ### Your templates and formatting Claude uses your rules, whether for slide masters, heading styles, formula structures, or numbering so you stay on brand. ### Enterprise security, built in Sign in with a Claude account, or connect through your existing cloud provider. Works within your existing compliance framework. How can Claude help? Open the brief in Word and start the memo from our firm template. Copy prompt [Copy prompt](https://claude.ai/new?q=Open+the+brief+in+Word+and+start+the+memo+from+our+firm+template.) Copy prompt How can Claude help? Turn this updated analysis into the SteerCo deck in the Acme template. Copy prompt [Copy prompt](https://claude.ai/new?q=Turn+this+updated+analysis+into+the+SteerCo+deck+in+the+Acme+template) Copy prompt How can Claude help? Find 30 minutes with everyone on this thread next week Copy prompt [Copy prompt](https://claude.ai/new?q=What+did+the+other+side+change+in+section+4%2C+and+are+any+of+these+dealbreakers%3F) Copy prompt [Prev](#) Prev Blog post Claude for Excel, PowerPoint, and Word are now generally available on all paid plans, with Claude for Outlook joining in beta. Read blog [Read blog](/blog/collaborate-with-claude-across-excel-powerpoint-word-and-outlook) Read blog ### What customers are saying “Claude enables a seamless transition from rough ideas to polished, branded deliverables across presentation, document and spreadsheet applications without disrupting the workflow. That end-to-end consistency has been a significant time-saver, especially when quality, formatting, and speed all matter.” Vivek Kulkarni, US AI Transformation Leader “Claude in Excel allows my teams to build an initial version of complex models faster, enabling them to focus on refining the model, pressure testing inputs and assumption, and exploring more scenarios and trade-offs. This helps us have richer and deeper discussions with our clients earlier.” Gene Rapoport, Head of Private Equity AI Practice “Claude for M365 is being rapidly adopted by ServiceNow. Claude does the work in Excel itself, instead of asking us to move content between tools, driving step change in productivity.” Rajeev Sethi, GVP Enterprise Technologies “We can instruct Claude to build a style guide from an executive's prior written communications so our EAs can draft their emails in their voice. That's high-leverage work the team would never have time to set up by hand.” Ben Letalik, Sr. Director, Digital Transformation & Innovation “Our investment professionals live in data and analytical models, and Claude for Excel meets them there. Analysts are using it to build and update coverage models, separate signal from noise, and pressure-test their work — all with a step-change in efficiency.” Atte Lahtiranta, Head of Core Engineering [Prev](#) Prev 0/5 ## Core capabilities across the apps you use every day ### One conversation across all your apps Context carries from Outlook to Word to Excel to PowerPoint. Change one thing, and every open file catches up. ### Deploy through your preferred cloud Connect through Microsoft Foundry, Amazon Bedrock, or Google Cloud Vertex AI credentials. ### Bring any document into the conversation Drop a PDF or doc into any sidebar. Claude reads it alongside your open file and references it in other apps. ### Turn workflows into Skills When a process is right, save it as a skill. The team can use it the same way in all four apps. Learn more [Learn more](/skills) Learn more ### Type or talk, whichever is faster Use dictation to talk to Claude in a sidebar instead of typing. ### Access connectors Pull context from outside sources directly from the sidebar. Learn more [Learn more](/connectors) Learn more ## FAQ ### Where is this available? Claude for Excel, PowerPoint, and Word are generally available on all paid Claude plans. Claude for Outlook is in beta on all paid plans. ### Can I connect through my cloud provider? Yes. Claude for Microsoft 365 is available on Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry. ### Does Claude preserve my formatting and templates? Yes. Claude inherits your heading styles, slide masters, formula conventions, and numbering. Edit one section without disturbing the rest. ### Can Claude send an email without me seeing it first? No. Drafts and calendar invites open in Outlook's native compose form and wait for you to click send. [Prev](#) Prev ## Get started with Claude for Microsoft 365 Install for Microsoft 365 [Install for Microsoft 365](https://marketplace.microsoft.com/en-us/product/office/WA200010725?tab=Overview) Install for Microsoft 365 Install for Outlook [Install for Outlook](https://marketplace.microsoft.com/en-us/product/office/WA200010724?tab=Overview) Install for Outlook [Homepage](https://claude.com) Homepage Thank you! Your submission has been received! Oops! Something went wrong while submitting the form. Write [Button Text](#) Button Text Learn [Button Text](#) Button Text Code [Button Text](#) Button Text Write - Help me develop a unique voice for an audience Hi Claude! Could you help me develop a unique voice for an audience? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Improve my writing style Hi Claude! Could you improve my writing style? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Brainstorm creative ideas Hi Claude! Could you brainstorm creative ideas? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! Learn - Explain a complex topic simply Hi Claude! Could you explain a complex topic simply? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Help me make sense of these ideas Hi Claude! Could you help me make sense of these ideas? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Prepare for an exam or interview Hi Claude! Could you prepare for an exam or interview? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! Code - Explain a programming concept Hi Claude! Could you explain a programming concept? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Look over my code and give me tips Hi Claude! Could you look over my code and give me tips? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Vibe code with me Hi Claude! Could you vibe code with me? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! More - Write case studies This is another test - Write grant proposals Hi Claude! Could you write grant proposals? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to — like Google Drive, web search, etc. — if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can - an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Write video scripts this is a test [Anthropic](https://www.anthropic.com/) Anthropic © \[year\] Anthropic PBC Products - Claude [Claude](/product/overview) Claude - Claude Code [Claude Code](/product/claude-code) Claude Code - Claude Code for Enterprise [Claude Code for Enterprise](/product/claude-code/enterprise) Claude Code for Enterprise - Claude Cowork [Claude Cowork](/product/cowork) Claude Cowork - @Claude [@Claude](/product/tag) @Claude - Claude Design [Claude Design](/product/design) Claude Design - Claude Science [Claude Science](/product/claude-science) Claude Science - Claude Security [Claude Security](/product/claude-security) Claude Security - Download app [Download app](/download) Download app - Pricing [Pricing](/pricing) Pricing - Log in [Log in](https://claude.ai/login) Features - Claude for Chrome [Claude for Chrome](/claude-for-chrome) Claude for Chrome - Claude for Microsoft 365 [Claude for Microsoft 365](/claude-for-microsoft-365) Claude for Microsoft 365 - Skills [Skills](/skills) Skills Models - Mythos [Mythos](https://www.anthropic.com/claude/mythos) Mythos - Fable [Fable](https://www.anthropic.com/claude/fable) Fable - Opus [Opus](https://www.anthropic.com/claude/opus) Opus - Sonnet [Sonnet](https://www.anthropic.com/claude/sonnet) Sonnet - Haiku [Haiku](https://www.anthropic.com/claude/haiku) Haiku Solutions - AI agents [AI agents](/solutions/agents) AI agents - Code modernization [Code modernization](/solutions/code-modernization) Code modernization - Coding [Coding](/solutions/coding) Coding - Customer support [Customer support](/solutions/customer-support) Customer support - Cybersecurity [Cybersecurity](/solutions/cybersecurity) Cybersecurity - Enterprise [Enterprise](/solutions/enterprise) Enterprise - Financial services [Financial services](/solutions/financial-services) Financial services - Government [Government](/solutions/government) Government - Healthcare [Healthcare](/solutions/healthcare) Healthcare - Higher education [Higher education](/solutions/education) Higher education - K-12 teachers [K-12 teachers](/solutions/teachers) K-12 teachers - Legal [Legal](/solutions/legal) Legal - Life sciences [Life sciences](/solutions/life-sciences) Life sciences - Nonprofits [Nonprofits](/solutions/nonprofits) Nonprofits - Small business [Small business](/solutions/small-business) Small business Claude Platform - Overview [Overview](/platform/api) Overview - Developer docs [Developer docs](https://platform.claude.com/docs) Developer docs - Pricing [Pricing](https://claude.com/pricing#api) Pricing - Ecosystem [Ecosystem](/ecosystem) Ecosystem - Marketplace [Marketplace](/platform/marketplace) Marketplace - Claude on AWS [Claude on AWS](/partners/claude-on-aws) Claude on AWS - Google Cloud [Google Cloud](/partners/google-cloud) Google Cloud - Microsoft Foundry [Microsoft Foundry](/partners/microsoft-foundry) Microsoft Foundry - Regional compliance [Regional compliance](/regional-compliance) Regional compliance - Console login [Console login](https://platform.claude.com/) Console login Resources - Blog [Blog](/blog) Blog - Claude partner network [Claude partner network](/partners) Claude partner network - Community [Community](/community) Community - Connectors [Connectors](/connectors) Connectors - Courses [Courses](https://www.anthropic.com/learn) Courses - Customer stories [Customer stories](/customers) Customer stories - Engineering at Anthropic [Engineering at Anthropic](https://www.anthropic.com/engineering) Engineering at Anthropic - Events [Events](https://www.anthropic.com/events) Events - Plugins [Plugins](/plugins) Plugins - Powered by Claude [Powered by Claude](/partners/powered-by-claude) Powered by Claude - Service partners [Service partners](/partners/services) Service partners - Tutorials [Tutorials](/resources/tutorials) Tutorials - Use cases [Use cases](/resources/use-cases) Use cases Company - Anthropic [Anthropic](https://www.anthropic.com/) Anthropic - Careers [Careers](https://www.anthropic.com/careers) Careers - Policy [Policy](https://www.anthropic.com/policy) Policy - Economic Futures [Economic Futures](https://www.anthropic.com/economic-futures) Economic Futures - Research [Research](https://www.anthropic.com/research) Research - News [News](https://www.anthropic.com/news) News - Policy on the AI Exponential --- ## claude-in-slack-tag-claude-in-any-thread-claude-by-anthropic-b8a28cc83f.md # Tag Claude in Slack [@Claude](https://anthropic.enterprise.slack.com/team/U08SSLN6TTL) reads threads, understands full context, and reacts in real time so your team moves forward together. Bring Claude into your channel. Add to Slack [Add to Slack](https://api.anthropic.com/integrations/v1/slack/install) Add to Slack Read documentation [Read documentation](https://claude.com/docs/claude-tag/overview) Read documentation Available in beta for Claude Enterprise and Team customers in Slack. [Play video](#) Play video “Claude Tag is our first responder for internal bugs. It reads incoming reports, looks at screenshots of failures, and uses access to Datadog, Linear, and GitHub to weed out user errors, trace root cause, and often draft a fix PR. Our engineers’ first reaction was ‘this is amazing.’” Aabhas Sharma, CTO “Claude Tag turned a summer of conference logistics — the follow-ups and chasing that always fell through the cracks — into an events program that runs itself while I stay on strategy.” Matthew Scott, Product Marketing “Claude Tag doesn’t just do the work — it challenges our thinking, provokes better questions, and elevates our judgement rather than replacing it. And we still feel very much in control throughout.” Simon Mansfield, Director, Enterprise AI “There were no skeptics on our team. Everyone leaned in from day one — it felt like Claude Tag in Slack was something they’d been waiting for.” Darian Bailey, AI Engineer & Tag Admin “Claude Tag meets our teams where they already work. Slack is the entry point for everything at Gusto, and the fact that Claude Tag is native there is huge for us. We see a lot of potential here.” Chad Kunsman, AI Developer [Prev](#) Prev 0/5 Coming soon to Microsoft Teams Tag [@Claude](https://anthropic.enterprise.slack.com/team/U08SSLN6TTL) in Teams, where it will read the conversation, understand context, and respond in real time. Join the waitlist [Join the waitlist](/form/claude-tag-teams-waitlist) Join the waitlist ## How it works Tag Claude into any thread in Slack, no matter how long or messy. It reacts to context, decisions, or open questions to help busy teams get more done. ### Tag it in @Claude in any thread or channel and it picks up the context everyone already shares. Ask it to run the numbers, file a ticket, or organize a chaotic chat into action items. It reads the room and does the work. Thread \#checkout-incidents × Marcus Nguyen 2:14 PM @Claude checkout error rate jumped from 0.3% to 6% right after the 14:00 deploy — can you dig in and figure out what broke? 1 reply ClaudeAGENT 2:14 PM On it — tracing the checkout error spike since the 14:00 deploy. ✓Pulled the 14:00 deploy diff from GitHub (`#4821`) ✓Queried Datadog — 502s isolated to the `cart-service` ✓Correlated with a 5× spike in DB connection-pool timeouts ✓Compared error rates by region against rollback candidates ✱Writing root-cause summary + rollback recommendation… To-dos as of 2:18 PM (Now) [Open session in Claude](#) [](#send) ### Set a schedule or let it run long tasks Give @Claude standing instructions and it works without a mention now or in the future. Ask it to watch a channel, run a weekly digest, flag anything urgent, or page the right person. \#revenue ⋮ Marcus Nguyen Thursday 8:55 AM @Claude Every Monday at 9, post a pipeline update for the team — pull the latest from Salesforce. ClaudeAGENT Thursday 8:55 AM Got it — I'll pull the pipeline from Salesforce and post a digest every Monday at 9 AM. Monday ▾ ClaudeAGENT 9:00 AM **Weekly pipeline — Jun 8–14** \$3.6M open pipeline · +\$610K WoW · 5 deals advanced. pipeline-trend.png▾ - Beacon Corp expansion advanced to **Negotiation** — \$240K - 2 new logos entered Discovery; 5 deals moved a stage - Acme renewal flagged at risk — pricing pushback, exec synced Thu [Open session in Claude](#) [](#threads) ### It jumps in, and can tag you back @Claude doesn't wait to be asked. It surfaces the thread that went quiet, posts when the deploy is live, and flags the decision that needs your call. Thread \#eng-triage × SentryAPP 10:02 AM **New issue** · `NullPointerException` in `CheckoutService.applyPromo()` 142 events in the last 30m · first seen 10:02 AM · 38 users affected [View in Sentry](#) 2 replies ClaudeAGENT 10:03 AM Reproduced — a null promo code slips past validation and hits `applyPromo()`. Fix is a one-line guard plus a regression test. ClaudeAGENT 10:05 AM Opened PR [\#1247](#) with the fix. cc @Marcus Nguyen — you own `checkout`, mind reviewing? [Open session in Claude](#) [](#get-help) ## How teams use @Claude Claude in Slack can share work in the format your team needs, right in the thread. Catch up Pull numbers Draft PRs Call prep Monitor channels Active Prompt @Claude What got decided here and what's still open? Search Workspace \#launch-q3 Message \#launch-q3 Pushing feature freeze to Thursday. Legal needs another pass on the data handling docs. Agreed. Also — pricing page copy is approved. Design can proceed. Can someone loop in Compliance on the pricing copy too, or is Legal's sign-off enough? @claude What got decided here and what's still open? ✓ Read through the channel from the top. ✓ Sort out what's settled vs. still pending. ✓ Write the summary. todos as of 1:22 PM [Open session in Claude](#) **Decided:** - Feature freeze moves to Thursday — Legal needs another pass on the data handling docs first. - Pricing page copy is approved — design can proceed. **Still open:** - No timeline yet on when Legal's pass will be done — Thursday freeze could slip again if it runs long. - Unclear who owns following up with Legal for status (Elena didn't assign it). [Open session in Claude](#) ### Catch up fast Get everything you need from a long thread in seconds. @Claude can surface decisions, open questions, and stakeholders, and you see exactly what's waiting on you. Prompt @Claude Top 20 enterprise accounts by spend, last 7 and 28 days. Search Workspace Thread \#revenue-ops × Reply… @Claude What are the top 20 enterprise accounts by spend, last 7 and 28 days? ✓ Pulled enterprise account spend from BigQuery for the trailing 7 and 28 days. ✓ Ranked the top 20 by 28-day spend and charted the 7-day alongside. todos as of 10:42 AM [Open session in Claude](#) Here are the top 20 enterprise accounts by spend, ranked by the trailing 28 days with the last 7 alongside. | Rank | Account | 7-day | 28-day | |------|-----------------------|--------|--------| | 1 | Northwind Trading | \$151K | \$512K | | 2 | Vertex Logistics | \$118K | \$438K | | 3 | Helios Energy | \$96K | \$391K | | 4 | Meridian Bank | \$88K | \$356K | | 5 | Atlas Robotics | \$101K | \$312K | | 6 | Sunbeam Health | \$64K | \$287K | | 7 | Cobalt Systems | \$71K | \$264K | | 8 | Orion Media | \$52K | \$241K | | 9 | Ironclad Security | \$70K | \$223K | | 10 | Pinnacle Foods | \$41K | \$205K | | 11 | Lumen Analytics | \$63K | \$188K | | 12 | Cascade Retail | \$38K | \$171K | | 13 | Granite Capital | \$44K | \$156K | | 14 | Beacon Aerospace | \$33K | \$142K | | 15 | Tideline Shipping | \$41K | \$129K | | 16 | Summit Pharma | \$24K | \$117K | | 17 | Verdant Agriculture | \$33K | \$104K | | 18 | Quartz Semiconductor | \$21K | \$92K | | 19 | Halcyon Games | \$29K | \$71K | | 20 | Driftwood Hospitality | \$14K | \$53K | top-20-enterprise-accounts-by-spend.png▾ The top 5 accounts are ~45% of enterprise spend (**\$2.0M** of **\$4.45M**). Worth watching **Atlas Robotics** and **Halcyon Games** — their last 7 days are 32% and 41% of the 28-day figure (a flat week would be ~25%), so both are accelerating. ### Pull the numbers Ask @Claude to query your data and it posts results like metrics, benchmarks, and charts in the channel, so the whole team sees the answer and can keep asking questions right there. Prompt @Claude Fix the bug in this thread and open a draft PR. Search Workspace \#eng-tickets Message \#eng-tickets Auth is throwing 502s on token refresh since the deploy. Looks like a race condition in `services/auth`. @Claude fix the bug in this thread and open a draft PR. Found it — race in `services/auth/token_cache.go`. Concurrent refreshes for the same subject both miss cache and the slower one overwrites the fresh token with a stale one → next request gets a 502. Deploy `a3f9c1e` exposed it by dropping the per-request mutex. Draft PR: [\#48217](#) · CI green · +71 −12 [Open session in Claude](#) ### Build from the thread Turn a bug report into a draft PR from the context is in the thread without ever leaving Slack. Get more done without switching tabs. Prompt @Claude I'm meeting Acme at 2 — what do I need to know? Search Workspace Thread \#acme-renewal × Reply… @Claude I’m meeting Acme at 2 — what do I need to know? **Acme · 2pm** — Sarah Chen (CTO, decision), Marcus Webb (VP Eng, champion) - **Deal:** \$1.4M renewal + expansion, close Jun 30. Redlined MSA with their legal since Fri. - **Sticking points:** EU data residency · 99.95% SLA ask · flat price on 62 overage seats - **Usage:** +38% MoM, 312/250 seats, 2 P1s last 30d (both \<4h) - **Watch:** “evaluating Cursor” thread in their \#eng-leads Tue — leverage, not real - **Play:** lead with usage + support SLAs → trade EU commit (Q3) for them dropping 99.95%. Let them raise the overage. [Deal room](#) · [last Gong recap](#) [Open session in Claude](#) ### Prep before calls Walk into every meeting with a full briefing. CRM notes, recent threads, and call history pulled together before you think to ask. Prompt @Claude Watch this channel, triage what comes in, and work through the backlog. Tag me only when something needs my call. Search Workspace \#prod-alerts Message \#prod-alerts @Claude Watch this channel, triage what comes in, and work through the backlog. Tag me only when something needs my call. On it — watching **\#prod-alerts** now. I’ll triage what comes in and work the backlog, and only tag you when something needs your call. 🚨 `checkout-api` — p99 latency **4,812ms** (threshold 2,000ms) env: `prod` · 3/12 hosts · started 02:06 UTC · [monitor](#) · [APM trace](#) On it. Spike lines up with deploy `c4f1a09` (02:04) — new inventory check in `CartService.validate()` is calling `inventory-svc` per line item instead of batched. Only the 3 canary hosts have it. ✓ Paused rollout at 25% ✓ Opened revert [\#48231](#) — merging once CI is green (~6m) ✱ Watching p99, should recover as canaries drain Not paging anyone. @Nathan — only flagging because the real fix (batch the inventory call) touches the pricing path you own; revert ships regardless. [Open session in Claude](#) ### Watch what matters @Claude can manage work in the channels your team never had bandwidth for: monitoring backlogs, triaging alerts, and surfacing what needs a human to review, so work that keeps getting pushed out can finally be completed. ## Core capabilities See what @Claude can do for your team. ### Understands your org The longer Claude works with your team, the more it understands how your organization thinks, how decisions get made, and who owns what. ### Memory across threads and days Context can draw on previous conversations and memory. What happened in Monday's standup is still there on Thursday, without anyone repeating it. ### A shared resource in your channel Claude surfaces what matters to the whole channel, not just the person who asked. Everyone sees the work, and anyone can build on it ### Proactive action instead of asking Set a standing instruction once. Claude watches the channel, runs routines, flags what's urgent, and tags you back when something needs a decision. ### In the work and tools with you Shape how Claude works in each channel with skills and instructions. Give it access to tools and data to take action on your behalf and post results. ### Runs tasks that take days Hand Claude a complex task and it works while you move on to other projects. It follows up, asks for input, or comes back when it's done. ## @Claude connects to your tools Configure any tool with an API using Agent Identity. Claude can query, act, and report back in the thread where the work is happening. Read more [Read more](/blog/agent-identity-access-model) Read more ## Admins stay in control Claude acts under its own identity in your systems. Admins configure access once; everyone else just types @Claude. ### **Customizable  identities for granular governance** Claude has its own account in your systems, not a borrowed login. Every credential’s use is logged, so "what did Claude do, and who asked for it?" always has an answer. ### Add it per channel Decide which tools Claude can reach and read. Confine sensitive connectors to a single private channel or enable broader access for specific tools and resources. Adjust any time from the admin console. What is Claude Tag? Learn more about [**@Claude**](https://anthropic.enterprise.slack.com/team/U08SSLN6TTL) and what happens to the former Claude for Slack experience. Read more [Read more](https://support.claude.com/en/articles/15594475) Read more Agent identity: a new security model for autonomous, team-wide AI Explore our new security model built for agents, not retrofitted from chatbots. Read more [Read more](/blog/agent-identity-access-model) Read more Tutorial: Working with @Claude in your workspace Claude now works alongside your team, under its own account, in the places you already work together. You tag it in the way you'd tag anyone, or have it speak up on its own when there's something it can help with. Read more [Read more](/resources/tutorials/best-practices-using-claude-tag) Read more [Prev](#) Prev ## Ready for a new way of working? Add to Slack [Add to Slack](https://api.anthropic.com/integrations/v1/slack/install) Add to Slack Join Teams waitlist [Join Teams waitlist](/form/claude-tag-teams-waitlist) Join Teams waitlist [Homepage](https://claude.com) Homepage Thank you! Your submission has been received! Oops! Something went wrong while submitting the form. Write [Button Text](#) Button Text Learn [Button Text](#) Button Text Code [Button Text](#) Button Text Write - Help me develop a unique voice for an audience Hi Claude! Could you help me develop a unique voice for an audience? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Improve my writing style Hi Claude! Could you improve my writing style? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Brainstorm creative ideas Hi Claude! Could you brainstorm creative ideas? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! Learn - Explain a complex topic simply Hi Claude! Could you explain a complex topic simply? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Help me make sense of these ideas Hi Claude! Could you help me make sense of these ideas? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Prepare for an exam or interview Hi Claude! Could you prepare for an exam or interview? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! Code - Explain a programming concept Hi Claude! Could you explain a programming concept? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Look over my code and give me tips Hi Claude! Could you look over my code and give me tips? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Vibe code with me Hi Claude! Could you vibe code with me? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! More - Write case studies This is another test - Write grant proposals Hi Claude! Could you write grant proposals? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to — like Google Drive, web search, etc. — if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can - an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Write video scripts this is a test [Anthropic](https://www.anthropic.com/) Anthropic © \[year\] Anthropic PBC Products - Claude [Claude](/product/overview) Claude - Claude Code [Claude Code](/product/claude-code) Claude Code - Claude Code for Enterprise [Claude Code for Enterprise](/product/claude-code/enterprise) Claude Code for Enterprise - Claude Cowork [Claude Cowork](/product/cowork) Claude Cowork - @Claude [@Claude](/product/tag) @Claude - Claude Design [Claude Design](/product/design) Claude Design - Claude Science [Claude Science](/product/claude-science) Claude Science - Claude Security [Claude Security](/product/claude-security) Claude Security - Download app [Download app](/download) Download app - Pricing [Pricing](/pricing) Pricing - Log in [Log in](https://claude.ai/login) Features - Claude for Chrome [Claude for Chrome](/claude-for-chrome) Claude for Chrome - Claude for Microsoft 365 [Claude for Microsoft 365](/claude-for-microsoft-365) Claude for Microsoft 365 - Skills [Skills](/skills) Skills Models - Mythos [Mythos](https://www.anthropic.com/claude/mythos) Mythos - Fable [Fable](https://www.anthropic.com/claude/fable) Fable - Opus [Opus](https://www.anthropic.com/claude/opus) Opus - Sonnet [Sonnet](https://www.anthropic.com/claude/sonnet) Sonnet - Haiku [Haiku](https://www.anthropic.com/claude/haiku) Haiku Solutions - AI agents [AI agents](/solutions/agents) AI agents - Code modernization [Code modernization](/solutions/code-modernization) Code modernization - Coding [Coding](/solutions/coding) Coding - Customer support [Customer support](/solutions/customer-support) Customer support - Cybersecurity [Cybersecurity](/solutions/cybersecurity) Cybersecurity - Enterprise [Enterprise](/solutions/enterprise) Enterprise - Financial services [Financial services](/solutions/financial-services) Financial services - Government [Government](/solutions/government) Government - Healthcare [Healthcare](/solutions/healthcare) Healthcare - Higher education [Higher education](/solutions/education) Higher education - K-12 teachers [K-12 teachers](/solutions/teachers) K-12 teachers - Legal [Legal](/solutions/legal) Legal - Life sciences [Life sciences](/solutions/life-sciences) Life sciences - Nonprofits [Nonprofits](/solutions/nonprofits) Nonprofits - Small business [Small business](/solutions/small-business) Small business Claude Platform - Overview [Overview](/platform/api) Overview - Developer docs [Developer docs](https://platform.claude.com/docs) Developer docs - Pricing [Pricing](https://claude.com/pricing#api) Pricing - Ecosystem [Ecosystem](/ecosystem) Ecosystem - Marketplace [Marketplace](/platform/marketplace) Marketplace - Claude on AWS [Claude on AWS](/partners/claude-on-aws) Claude on AWS - Google Cloud [Google Cloud](/partners/google-cloud) Google Cloud - Microsoft Foundry [Microsoft Foundry](/partners/microsoft-foundry) Microsoft Foundry - Regional compliance [Regional compliance](/regional-compliance) Regional compliance - Console login [Console login](https://platform.claude.com/) Console login Resources - Blog [Blog](/blog) Blog - Claude partner network [Claude partner network](/partners) Claude partner network - Community [Community](/community) Community - Connectors [Connectors](/connectors) Connectors - Courses [Courses](https://www.anthropic.com/learn) Courses - Customer stories [Customer stories](/customers) Customer stories - Engineering at Anthropic [Engineering at Anthropic](https://www.anthropic.com/engineering) Engineering at Anthropic - Events [Events](https://www.anthropic.com/events) Events - Plugins [Plugins](/plugins) Plugins - Powered by Claude [Powered by Claude](/partners/powered-by-claude) Powered by Claude - Service partners [Service partners](/partners/services) Service partners - Tutorials [Tutorials](/resources/tutorials) Tutorials - Use cases [Use cases](/resources/use-cases) Use cases Company - Anthropic [Anthropic](https://www.anthropic.com/) Anthropic - Careers [Careers](https://www.anthropic.com/careers) Careers - Policy [Policy](https://www.anthropic.com/policy) Policy - Economic Futures [Economic Futures](https://www.anthropic.com/economic-futures) Economic Futures - Research [Research](https://www.anthropic.com/research) Research - News [News](https://www.anthropic.com/news) News - Policy on the AI Exponential --- ## enabling-and-using-the-microsoft-365-connector-f39b429330.md # Set up the Microsoft 365 connector This article walks admins through enabling the Microsoft 365 connector for their organization in Claude—including granting Microsoft Entra consent, restricting access, and managing permissions. Once setup is complete, people in your tenant can connect Microsoft 365 to their own Claude accounts to search across SharePoint, OneDrive, Outlook, and Teams from Claude. You can also enable write tools, which let Claude send email, manage calendar events, and create and update files on a member's behalf. The Microsoft 365 connector is available on all Claude plans: Free, Pro, Max, Team, and Enterprise. For end-user instructions on connecting and using Microsoft 365 once setup is complete, see **[Connect Claude to Microsoft 365](https://support.claude.com/en/articles/15183774-)**. **Important:** The Microsoft 365 connector requires a Microsoft Entra tenant tied to a Microsoft Business plan. Personal Microsoft accounts (such as @outlook.com or @hotmail.com addresses) can't be used to connect. ------------------------------------------------------------------------ ## Setup overview Two things need to happen before anyone in your organization can connect Microsoft 365, plus an optional third step if you want to enable write tools: 1. **On Team and Enterprise plans:** A Claude organization owner enables the Microsoft 365 connector for the organization. 2. **In every tenant:** A Microsoft Entra Global Administrator grants a one-time consent that authorizes the integration for your tenant. 3. **To enable write tools:** A Microsoft Entra administrator consents to the updated permission set, and you enable write tools for your organization. See **[Enable write tools](#h_a51d877afd)** below. After completing these steps, members can connect Microsoft 365 to their own Claude accounts following the steps in **[Connect Claude to Microsoft 365](https://support.claude.com/en/articles/15183774-)**. ## Enable the connector for your organization This step applies to Team and Enterprise plans only. On Free, Pro, and Max plans, skip to the next section. 1. Sign in to Claude. 2. Navigate to **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. 3. Click “+ Add” at the top of the page, then “All available.” 4. Find **Microsoft 365** and click “Add to your team.” ## Grant Microsoft Entra admin consent A Microsoft Entra Global Administrator in your tenant needs to authorize the integration before anyone can connect. There are two ways to do this. ### Option 1: Consent through Claude If your Microsoft Entra Global Administrator has a Claude account, they can grant consent during the standard connection flow: 1. Navigate to **[Customize \> Connectors](http://claude.ai/customize/connectors)**. 2. Find **Microsoft 365** and click “Connect.” 3. Authenticate with Microsoft 365 credentials. 4. Review and accept the requested permissions, checking the box to grant access on behalf of the whole organization. After this, other people in the same Entra tenant can connect by following the standard end-user steps. They won't see the consent prompt—they'll just authenticate and start using the integration. ### Option 2: Manual setup in Microsoft Entra ID Use this path if your Microsoft Entra Global Administrator doesn't have a Claude account, or if you need to troubleshoot the app install and permissions setup. You can add the connector apps and grant admin consent directly in Microsoft Entra ID. This process adds two service principals to your tenant. Each principal establishes a service-level identity for one of the two M365 MCP for Claude app registrations, allowing them to access and interact with your organization's data and resources via the Microsoft Graph API. **1. Add the service principals** Using Microsoft Graph Explorer, add both required service principals: M365 MCP Client for Claude: POST https://graph.microsoft.com/v1.0/servicePrincipals {"appId":"08ad6f98-a4f8-4635-bb8d-f1a3044760f0"} M365 MCP Server for Claude: POST https://graph.microsoft.com/v1.0/servicePrincipals {"appId":"07c030f6-5743-41b7-ba00-0a6e85f37c17"} **2. Grant admin consent** Construct and visit the following URLs in your browser, replacing {your-tenant-id} with your organization's tenant ID. M365 MCP Client for Claude: https://login.microsoftonline.com/{your-tenant-id}/adminconsent?client_id=08ad6f98-a4f8-4635-bb8d-f1a3044760f0 M365 MCP Server for Claude: https://login.microsoftonline.com/{your-tenant-id}/adminconsent?client_id=07c030f6-5743-41b7-ba00-0a6e85f37c17 When you visit each URL, you'll be prompted to consent to the delegated permissions required by the integration on behalf of your organization. **3. Finish setup** - **Team and Enterprise plans:** A Claude organization Owner needs to enable the connector in **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. Then members can connect individually. - **Free, Pro, and Max plans:** Members can connect by navigating to **[Customize \> Connectors](http://claude.ai/customize/connectors)**, finding **Microsoft 365**, and clicking “Connect.” ### Restrict who can use the connector To limit which people in your tenant can authenticate to Microsoft 365 through Claude: 1. Go to the Microsoft Entra admin center at entra.microsoft.com. 2. Navigate to the **M365 MCP Server for Claude** enterprise application. 3. Go to **Properties** and set **Assignment required?** to “Yes.” 4. Under **Users and groups**, add the specific users or groups who should have access. 5. Repeat the same steps for the **M365 MCP Client for Claude** enterprise application. Both components need to be restricted to the same set of authorized people. ### Restrict which permissions the connector can use To limit which types of resources the integration can access, selectively revoke permissions from the default set of authorized scopes. This requires Microsoft Entra admin access. 1. As a Microsoft Entra admin, go to entra.admin.com. 2. Select “Enterprise Applications.” 3. Next to the search box, remove the application type filter. 4. Search for and click “M365 MCP Server for Claude.” 5. Go to **Permissions**. 6. Under the **Admin consent** tab and in the Microsoft Graph list of permissions, select the permission you would like to revoke and click the “**…**” button. 7. Select “Revoke permission” and confirm with “Yes, revoke.” Once revoked, attempts to access a resource with that permission will return a "Failed to call tool" error. Members can also individually turn off specific tools in their own Microsoft 365 settings to prevent Claude from trying to access a tool for which the permission has been revoked. To restore a revoked permission, follow the steps to grant admin consent described in **Option 2: Manual setup in Microsoft Entra ID**. This will revert the permissions to the default state. ------------------------------------------------------------------------ ## Enable write tools Write tools let Claude send email, manage drafts and calendar events, update mailbox settings, and create and update files in OneDrive and SharePoint. Read and search tools work the same whether or not write tools are enabled. **1. Re-consent to the updated permissions** The connector's permission set now includes additional Microsoft Graph scopes to support write tools. If your tenant consented before write tools launched, a Microsoft Entra Global Administrator needs to review and approve the updated permission set before write tools activate. Review and approve the updated permissions for the connector in your tenant's **Enterprise Applications** consent flow. This is a one-time action per tenant. **2. Enable write tools for your organization** If your organization was using the connector before write tools launched, they will be blocked by default. Enable them for everyone by going to **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**, finding “Microsoft 365,” and setting the appropriate permissions. Enterprise plans can enable them for a subset of users through **[custom roles](https://support.claude.com/en/articles/13930452-manage-custom-roles-on-enterprise-plans#h_979e558d00)**. **3. Verify** Once enabled, ask Claude to perform a low-risk write action, such as "Draft an email to myself, but don't send it," to confirm write tools are active. **Note:** Emails Claude sends include an attribution header identifying them as agent-initiated. File and calendar writes aren't currently tagged. Attachments aren’t supported in write tools, so sending, forwarding, and drafting all reject messages with attachments. Write tools are also subject to per-user limits on writes, sends, and recipients. ------------------------------------------------------------------------ ## Permissions reference The Microsoft 365 connector uses **delegated permissions**, meaning Claude acts on behalf of each individual user and can only access data that user already has permission to view in Microsoft 365. Permissions are read-only—Claude can't modify, delete, or create content in your tenant. During authentication, the integration requests the following permissions: **Basic access** - `User.Read`: Sign in and read user profile - `openid`: Sign in with organizational account - `offline_access`: Maintain access to data - `email`: View email address - `profile`: View basic profile information **Email (Outlook)** - `Mail.Read`: Read email messages - `Mail.ReadBasic`: Read email metadata (sender, subject, date) - `Mail.Read.Shared`: Read emails in mailboxes the user has access to - `MailboxFolder.Read`: Read mailbox folder structure - `MailboxItem.Read`: Read items in mailbox - `MailboxSettings.Read`: Read mailbox settings, like the user's timezone **Note:** Shared mailbox access is included through the `Mail.Read.Shared` permission. Members can search shared mailboxes they have delegate access to in Microsoft 365, including full access and folder-level delegation. No setup is needed beyond standard admin consent and the delegate permissions already configured in Microsoft 365. **Calendar** - `Calendars.Read`: Read calendar events - `Calendars.Read.Shared`: Read calendars shared with the user **Teams chat** - `Chat.Read`: Read Teams chat messages - `Chat.ReadBasic`: Read Teams chat metadata - `ChatMember.Read`: Read information about chat participants - `ChatMessage.Read`: Read Teams chat messages **Teams channels** - `Channel.ReadBasic.All`: Read channel names and descriptions - `ChannelMessage.Read.All`: Read channel messages **Meetings** - `OnlineMeetings.Read`: Read online meetings - `OnlineMeetingTranscript.Read.All`: Read meeting transcripts - `OnlineMeetingAiInsight.Read`: Read AI-generated meeting insights - `OnlineMeetingArtifact.Read.All`: Read meeting recordings and artifacts - `OnlineMeetingRecording.Read.All`: Read meeting recordings **Files (OneDrive and SharePoint)** - `Files.Read`: Read user files - `Files.Read.All`: Read all files the user can access - `Sites.Read.All`: Read items in SharePoint sites **Write permissions** The following permissions support write tools and are included in the updated consent set: - `Mail.Send`: Send and forward email - `Mail.ReadWrite`: Create, update, and delete drafts; move and label messages - `Calendars.ReadWrite`: Create, update, delete, and respond to calendar events - `Files.ReadWrite.All`: Create and update files in OneDrive and SharePoint - `MailboxSettings.ReadWrite`: Manage categories, inbox rules, and automatic replies **User directory** - `User.ReadBasic.All`: Read basic profile information for all users in the organization (used for finding meeting availability) The Microsoft 365 connector searches SharePoint across the entire tenant using the permissions of the user. Site-specific search restriction isn't supported. ## Privacy and security - **Permission inheritance:** Claude mirrors each user's existing Microsoft 365 permissions. Members can't access anything through Claude that they couldn't already see directly in Microsoft 365. - **On-demand access:** Claude only accesses data when a user explicitly asks a question that requires it. - **Revocable access:** Members can disconnect their own integration through **[Customize \> Connectors](http://claude.ai/customize/connectors)**. Team and Enterprise plan Owners can also remove the connector for the entire organization in **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. For more detail, see the **[Microsoft 365 connector security guide](https://support.claude.com/en/articles/12684923-)**. ------------------------------------------------------------------------ ## Troubleshooting ### A member can't authenticate 1. Confirm their account is tied to a Microsoft Entra tenant, not a personal Microsoft account. 2. Confirm their Microsoft 365 license is active. 3. Confirm admin consent has been granted using Option 1 or Option 2 above. 4. Check whether organizational policies (such as conditional access) are blocking third-party app authentication. ### Members are seeing "Failed to call tool" errors A permission may have been selectively revoked in Microsoft Entra. Members can turn off the corresponding tool in their Microsoft 365 settings to suppress the error, or you can restore the permission by repeating the admin consent steps in **[Option 2: Manual setup in Microsoft Entra ID](#h_c5f095c9cf)**. ### Write tools aren't appearing for members 1. Confirm a Microsoft Entra administrator has consented to the updated permission set that includes write scopes. 2. Confirm write tools are enabled in the Microsoft 365 connector configuration, or that the member is covered by a role-based access policy that grants them. 3. Have the member disconnect and reconnect Microsoft 365 in **[Customize \> Connectors](http://claude.ai/customize/connectors)**. ------------------------------------------------------------------------ ## Frequently asked questions ### What happens if a member tries to connect before consent is granted? They'll see an error message indicating that an administrator must grant app permissions before they can use the integration. The connection will fail until a Microsoft Entra Global Administrator approves the necessary permissions. ### Can the Microsoft 365 connector be used with enterprise search? Yes. When enterprise search is enabled, it can query Microsoft 365 alongside other connected services for unified search across Slack, Google Workspace, Microsoft 365, and more. ### Can the integration modify Microsoft 365 data? Only after an Entra admin grants write scopes. With write tools on, Claude can send email, manage drafts and calendar events, update mailbox settings, and create and update files in OneDrive and SharePoint, always within each member's existing Microsoft 365 permissions. Without them, the integration is read-only. Claude can't post Teams messages or change Teams settings or permissions in either case, since there are no tools allowing this. --- ## enabling-and-using-the-microsoft-365-connector.md 4. Enabling and using the Microsoft 365 connector # Enabling and using the Microsoft 365 connector The Microsoft 365 connector is available for users on Team and Enterprise plans. This article explains how to connect Claude to Microsoft 365 using our pre-built MCP connector, allowing Claude to search, analyze, and access information across SharePoint, OneDrive, Outlook, and Teams. With Microsoft 365 connected, Claude can: - **Search and analyze documents** across SharePoint sites and OneDrive libraries - **Access email threads** and analyze communications from Outlook - **Review meeting information** from Teams Calendar - **Pull insights** from Teams Chat discussions ## Enabling the Microsoft 365 connector Enabling the Microsoft 365 connector for your organization requires two separate setup phases with specific steps that must be completed by a Microsoft Entra ID Global Administrator, and a Claude Team or Enterprise plan Owner. **Prerequisites for enablement:** - A Claude user with an Owner or Primary Owner role on a Team or Enterprise organization - Someone with Global Administrator access to your organization's Microsoft Entra tenant - Users must have Microsoft 365 accounts to connect to the connector and start using it with Claude. ### Phase 1: Initial Microsoft Entra Global Administrator Setup A Microsoft Entra Global Administrator must complete a one-time setup process before Claude Team and Enterprise plan users can connect to the Microsoft 365 connector. **Automatic Setup through Auth Consent Flow (Recommended)** This process triggers an auth consent flow when connecting to the Microsoft 365 connector for the first time as a Global Administrator. **Steps:** **1) Enable in Claude organization settings** As an organization Owner and Microsoft Entra Global Administrator: 1. Sign in to Claude. 2. Navigate to **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. 3. Click the “Browse connectors” button at the bottom of the page. 4. Find "Microsoft 365" and click "Add to your team." “Microsoft 365” will now appear in the list of Connectors at **[Settings \> Connectors](https://claude.ai/settings/connectors)**. **2) Enable in individual Claude Settings** After completing the previous step on behalf of the Claude organization, the Microsoft Entra Global Administrator needs to connect to Microsoft 365 in their individual Claude user settings: 1. Navigate to **[Settings \> Connectors](https://claude.ai/settings/connectors)**. 2. Find "Microsoft 365" in the list and click "Connect." 3. Authenticate with your Microsoft 365 credentials. 4. You can then review and accept the requested permissions, checking the box to grant access on behalf of the whole organization. - You must complete this step before any team members can connect to Microsoft 365 individually. 5. (Optional) **To restrict which users in your Microsoft Entra tenant can use the connector**, navigate to the M365 MCP Server for Claude enterprise application in the Entra admin center (**[https://entra.microsoft.com](https://entra.microsoft.com)**), go to Properties and set “Assignment required?” to Yes, then add the specific users or groups under the Users and groups section—only those assigned will be able to authenticate and use the connector. Repeat this same process for the M365 MCP Client for Claude enterprise application to ensure both components are restricted to the same set of authorized users. 6. (Optional) **To selectively restrict which permissions scopes users in your tenant can use** see the below “Permission categories” and “Selectively revoking permissions” sections. **Manual Setup in Microsoft Entra ID** Alternatively, you can add the connector apps and grant admin pre-consent on behalf of the whole tenant in Microsoft Entra ID. This process manually achieves what is done above in the Automatic Setup through Auth Consent Flow section above. You can use this method if your Microsoft Entra Global Administrator is not a member of the Claude Team or Enterprise organization or to troubleshoot the app install and permissions setup in Microsoft Entra ID. This process adds two service principals to Graph Explorer; each principal establishes a service-level identity for one of the two M365 MCP for Claude app registrations in your tenant, allowing them to access and interact with your organization's data and resources via the Microsoft Graph API. **Steps:** **1) Add the service principals** Using [Microsoft Graph Explorer](https://developer.microsoft.com/en-us/graph/graph-explorer), add both required service principals: M365 MCP Client for Claude POST [https://graph.microsoft.com/v1.0/servicePrincipals](https://graph.microsoft.com/v1.0/servicePrincipals) {"appId":"08ad6f98-a4f8-4635-bb8d-f1a3044760f0"} M365 MCP Server for Claude POST [https://graph.microsoft.com/v1.0/servicePrincipals](https://graph.microsoft.com/v1.0/servicePrincipals) {"appId":"07c030f6-5743-41b7-ba00-0a6e85f37c17"} **2) Grant admin pre-consent** Construct and visit the following URLs in your browser, replacing {your-tenant-id} with your organization's tenant ID: **M365 MCP Client for Claude** [https://login.microsoftonline.com/{your-tenant-id}/adminconsent?client_id=08ad6f98-a4f8-4635-bb8d-f1a3044760f0](https://login.microsoftonline.com/%7Byour-tenant-id%7D/adminconsent?client_id=08ad6f98-a4f8-4635-bb8d-f1a3044760f0) **M365 MCP Server for Claude** [https://login.microsoftonline.com/{your-tenant-id}/adminconsent?client_id=07c030f6-5743-41b7-ba00-0a6e85f37c17](https://login.microsoftonline.com/%7Byour-tenant-id%7D/adminconsent?client_id=07c030f6-5743-41b7-ba00-0a6e85f37c17) When you visit each URL, you'll be prompted to consent to the delegated user permissions required by the integration on behalf of your organization. **3) Enable in Claude organization settings** After the Microsoft Entra Admin completes the admin consent process, an organization Owner needs to: 1. Sign in to Claude. 2. Navigate to **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. 3. Click the “Browse connectors” button at the bottom of the page. 4. Find "Microsoft 365" and click "Add to your team." “Microsoft 365” will now appear in the list of Connectors at **[Settings \> Connectors](https://claude.ai/settings/connectors)**. **4) Enable in individual Claude Settings** After completing the previous step on behalf of the organization, the Microsoft Entra Admin needs to connect to Microsoft 365 in their individual Claude user settings: 1. Navigate to **[Settings \> Connectors](https://claude.ai/settings/connectors)**. 2. Find "Microsoft 365" in the list and click "Connect." 3. Authenticate with your Microsoft 365 credentials. 4. You can then review and accept the requested permissions, checking the box to grant access on behalf of the whole organization. - You must complete this step before any team members can connect to Microsoft 365 individually. 5. (Optional) **To restrict which users in your Microsoft Entra tenant can use the connector**, navigate to the M365 MCP Server for Claude enterprise application in the Entra admin center (**[https://entra.microsoft.com](https://entra.microsoft.com)**), go to Properties and set “Assignment required?” to Yes, then add the specific users or groups under the Users and groups section—only those assigned will be able to authenticate and use the connector. Repeat this same process for the M365 MCP Client for Claude enterprise application to ensure both components are restricted to the same set of authorized users. 6. (Optional) **To selectively restrict which permissions scopes users in your tenant can use** see the below “Permission categories” and “Selectively revoking permissions” sections. **5) Test the connector** After you connect successfully, start a new chat with Claude and try making a simple request (e.g., “List all of my SharePoint docs”). If Claude can access the requested data using the Microsoft 365 connection, this confirms the connector is working. All Claude organization members can start authenticating with and using the Microsoft 365 connector. ### Phase 2: Enablement steps for Team and Enterprise users Once enabled by an Owner, members can choose to connect Claude to Microsoft 365 in their settings. **Steps:** 1. Navigate to **[Settings \> Connectors](https://claude.ai/settings/connectors)**. 2. Find "Microsoft 365" in the list and click "Connect." 3. Authenticate with your Microsoft 365 credentials to start using the Microsoft 365 connector with Claude. **Note:** Once you've added the Microsoft connector to your Claude account, you can authenticate with the tool and use it on Claude for iOS or Android. ------------------------------------------------------------------------ ## How to use the Microsoft 365 connector Ask Claude a question that requires accessing your Microsoft 365 data. Claude will automatically detect which tools it needs and retrieve the relevant information. ### Example queries - "Find the Q4 strategic planning document in SharePoint." - "Summarize email conversations about the product launch." - "What discussions happened in the Teams channel about the marketing campaign?" - "Review meeting notes from last week's leadership sync." Claude will provide a response based on information retrieved from your Microsoft 365 environment, including relevant context and citations when applicable. ## SharePoint and OneDrive document access - Search documents across SharePoint sites and libraries to locate project specifications, strategic plans, and other business documents. - Access files stored in your OneDrive and have Claude analyze content without manually uploading. - Consolidate information from distributed file locations and analyze trends across multiple documents. ## Outlook email analysis - Search email threads and conversations to track project status, client feedback, and team alignment. - Access message content and metadata, filtering by date, sender, subject, and other criteria. - Analyze communication patterns and find specific information from past correspondences. ## Outlook Calendar meeting analysis and summarization - Review meeting summaries, attendee information, and content to prepare for upcoming meetings or understand discussions you missed. - Analyze scheduling patterns and track project decisions. ## Teams chat capabilities - Access Teams chat messages and channel discussions where you're a participant. - Review team collaboration patterns and find decisions made across conversations.\ ------------------------------------------------------------------------ ## Which permissions does the Microsoft 365 Connector require? When you connect the Microsoft 365 integration, you'll be asked to grant several permissions that allow Claude to access your Microsoft 365 data on your behalf. **Important to understand:** - All permissions are **delegated permissions**, which means Claude acts on behalf of the user’s Microsoft 365 account and can only access data they already have permission to view in Microsoft 365. - Claude can only access Microsoft 365 data for the account you've connected. - Claude cannot access anything beyond your existing permissions. - These permissions enable read-only access—Claude cannot modify, delete, or create content in your Microsoft 365 tenant. ## Permission categories During authentication, the Microsoft 365 connector requests the following permissions: **Basic access** - **User.Read**: Sign in and read your user profile - **openid: Sign in with your organizational account** - **offline_access: Maintain access to data you have given it access to** - **email: View your email address** - **profile: View your basic profile information** **Email (Outlook)** - **Mail.Read**: Read your email messages - **Mail.ReadBasic**: Read email metadata (sender, subject, date) - **Mail.Read.Shared**: Read emails in mailboxes you have access to - **MailboxFolder.Read**: Read your mailbox folder structure - **MailboxItem.Read**: Read items in your mailbox **Calendar** - **Calendars.Read**: Read your calendar events - **Calendars.Read.Shared**: Read calendars shared with you **Teams Chat** - **Chat.Read**: Read your Teams chat messages - **Chat.ReadBasic**: Read Teams chat metadata - **ChatMember.Read**: Read information about chat participants - **ChatMessage.Read**: Read your Teams chat messages **Teams Channels** - **Channel.ReadBasic.All**: Read channel names and descriptions - **ChannelMessage.Read.All**: Read channel messages **Meetings** - **OnlineMeetings.Read**: Read your online meetings - **OnlineMeetingTranscript.Read.All**: Read meeting transcripts - **OnlineMeetingAiInsight.Read**: Read AI-generated meeting insights - **OnlineMeetingArtifact.Read.All**: Read meeting recordings and artifacts - **OnlineMeetingRecording.Read.All**: Read meeting recordings **Files (OneDrive and SharePoint)** - **Files.Read**: Read your files - **Files.Read.All**: Read all files you can access - **Sites.Read.All**: Read items in SharePoint sites **User Directory** - **User.ReadBasic.All**: Read basic profile information for all users in your organization (used for finding meeting availability) ## Why are these permissions needed? These permissions allow Claude to do the following when prompted: - Search your emails, documents, and calendar to answer your questions. - Access meeting information and Teams discussions. - Find and analyze content across your Microsoft 365 environment. - Provide accurate, contextual responses based on your work data. Additionally, the Microsoft 365 Connector searches SharePoint across the entire tenant using the permissions of the user. Site-specific search restriction is unsupported. You can revoke these permissions as a user at any time by disconnecting the connector in your Claude settings, or as an organization by removing the connector in your Claude organization settings. ## Selectively revoking permissions To limit which types of resources the connector is able to access, you can selectively revoke permissions from the default set of authorized scopes the connector uses to access the Microsoft Graph API. 1. As a Microsoft Entra Admin, go to: entra.admin.com 2. Select “Enterprise Applications.” 3. Next to the search box, remove the application type filter. 4. Search for and click "M365 MCP Server for Claude." 5. Go to Permissions. 6. Under the Admin consent tab and in the Microsoft Graph list of permissions, select the permission you would like to revoke and click the breadcrumbs button (“...”). 7. Select “Revoke permission,” and confirm with the “Yes, revoke” button. 8. Claude will now be unable to access resources via that API. Attempts to access a resource with a revoked permission will show a "Failed to call tool \". 9. As a convenience, users can also individually toggle off which tools the connector will use in the Microsoft 365 connector settings to prevent Claude from trying to access a tool for which the permission is revoked. To restore a revoked permission, follow the steps to grant admin pre-consent described in **[Phase 1: Initial Microsoft Admin Setup](#h_f93cbae211)**. This will revert the permissions to the default state. ## Privacy and security - **Permission inheritance:** Claude mirrors your existing Microsoft 365 permissions. - **On-demand access:** Claude only accesses your data when you explicitly ask questions requiring it. - **Revocable access:** You can disconnect the integration at any time through **[Settings \> Connectors](https://claude.ai/settings/integrations)**. Read more here: **[Microsoft 365 Connector: Security Guide](microsoft-365-connector-security-guide.md)**.\ ------------------------------------------------------------------------ ## Troubleshooting ### Authentication is failing. What should I check? 1. **Verify your credentials:** Ensure you're using the correct Microsoft 365 account. 2. **Check subscription status:** Confirm your Microsoft 365 license is active. 3. **Review organizational policies:** Your IT team may need to approve third-party app access. 4. **Try a different browser:** Some browsers may block authentication popups. 5. **Disable browser extensions:** Ad blockers or privacy extensions may interfere. 6. **Clear cookies and cache:** Try a fresh browser session. ### Claude says it can't find documents I know exist Check the following: 1. **Permissions:** Verify you have access to the document in Microsoft 365 directly. 2. **Location:** Ensure the document is in SharePoint or OneDrive, not local storage. 3. **Indexing delay:** Recently uploaded documents may take time to become searchable. 4. **Specific location:** Try specifying the exact SharePoint site or library name. 5. **File name:** Try searching by exact file name or unique keywords from the document. ### Search results are incomplete or irrelevant Tips to improve your search queries: - Be more specific about what you're looking for. - Specify locations (site names, date ranges, document types). - Use exact phrases for better matching. - Try breaking up complex queries into simpler, more focused questions. - Verify spelling of names, projects, or terms.\ ------------------------------------------------------------------------ ## Frequently asked questions ### Can Claude modify my Microsoft 365 data? No. The current Microsoft 365 integration provides **read-only access**. Claude can search and analyze your data but cannot: - Create, edit, or delete documents - Send emails or calendar invites - Modify SharePoint sites or OneDrive files - Change Teams settings or permissions ### Can I use the Microsoft 365 Connector with Enterprise Search? Yes, the Microsoft 365 Connector works well with **[Enterprise Search](../13-Enterprise-Admin/using-enterprise-search.md)**. When enabled: 1. Enterprise Search can query Microsoft 365 alongside other connected tools. 2. You get unified search across Slack, Google Workspace, Microsoft 365, and more. 3. Enterprise Search's optimized prompts help Claude search more effectively. ### Can Claude search archived emails? Yes, Claude can search all emails you have access to in Outlook, including archived messages, as long as they're accessible through your account. ### Does Claude search shared drives and team sites? Yes, Claude can search any SharePoint sites and shared drives you have permission to access. This includes: - Team sites - Communication sites - SharePoint document libraries - Shared OneDrive folders ### Can Claude access private Teams channels? Claude can only access Teams content that you have permission to view in Microsoft 365. If you're a member of a private channel, Claude can search for that content. If you're not a member, Claude cannot access it. ### How do I ask Claude to search specific locations? Be specific in your queries: - **For specific SharePoint sites:** "Search the Engineering team site in SharePoint for architecture documents." - **For specific date ranges:** "Find emails from the last week about the Q4 budget." - **For specific senders or topics:** "Show me Teams discussions with Sarah about the product roadmap." - **For specific file types:** "Find PowerPoint presentations in SharePoint about sales strategy." ### Can Claude summarize long email threads? Yes. One of Claude's strengths is analyzing and summarizing complex communications. Try: "Summarize the email thread about the vendor selection process." Claude will read the thread and provide a concise summary with key points and decisions. ### What happens if a Microsoft 365 user tries to connect before a MicrosoftEntra Global Admininstrator grants tenant-wide permission? If a user without Microsoft Entra Global Administrator permissions attempts to connect their Microsoft 365 account, they will receive an error message indicating that an Administrator must grant app permissions before they can use the connector. The connection attempt will fail until a Microsoft Entra Global Administrator approves the necessary permissions as detailed above in Phase 1. --- ## get-started-with-claude-compliance-api-integrations-6e6376c9e5.md # Get started with Claude Compliance API integrations Security and compliance platforms have built integrations on top of the Claude Compliance API, so your team can monitor Claude activity within the tools you already use. Integrations are available from security and compliance providers across the categories enterprises rely on, spanning DLP, SASE, data security, SIEM and security operations, identity, eDiscovery, AI security posture management, and AI observability and telemetry infrastructure. This article lists available Compliance API integrations with a link to their setup documentation. If you're a security, compliance, or IT platform that has built a Compliance API integration, **[apply here to join the program](https://forms.gle/1rkBcdwwR5bLLD82A)**. The Claude Compliance API is available to Claude Enterprise plans and Claude Platform customers. ## What data you can access The Compliance API provides access to different data depending on the product: - **Claude Enterprise:** - **Conversation content** including chats, uploaded files, and projects - **Activity feed events** covering user logins, admin actions, and configuration changes - **Claude Platform:** - **Activity feed events**, including admin and system events (such as member and workspace changes, API key creation, and account settings) and resource events (such as file creation, downloads, and skill changes) - **Note:** Conversation content (prompts and model responses) is not available through the Compliance API for the Claude Platform. ## How to set up an integration For Claude Enterprise customers, using an integration is straightforward: 1. Enable the Compliance API in your organization settings. Learn how to **[access the Compliance API](https://support.claude.com/en/articles/13015708-access-the-compliance-api)**. 2. Connect your instance to a supported security platform. Setup guides are available in the **[Available integrations section](#h_7ff9e34f8e)**. Claude activity will flow into the same dashboards and workflows your team uses for every other application. For Claude Platform, contact your Anthropic sales team. Review **[Compliance API documentation](https://platform.claude.com/docs/en/manage-claude/compliance-api)** on Claude API Docs. ## Available integrations - **Air:** Air utilizes the Claude Compliance API as a core part of its agentic security platform, giving security teams unified posture management, real-time runtime protection, and supply-chain governance across all of their endpoint, cloud, and SaaS agents from a single control plane. To set up this integration, see the **[Air setup guide](https://www.air.security/claude-air-integration)**. - **Airia:** Airia's integration with the Compliance API gives security and risk teams policy enforcement, automated compliance reporting, and continuous audit trails across Claude usage, so enterprises can detect risky behavior, enforce governance policies, and demonstrate regulatory compliance with confidence. To set up this integration, see the **[Airia setup guide](https://explore.airia.com/integrations/Tools/anthropic-compliance-api)**. - **aizome:** aizome discovers and inventories how Claude is used across your organization—users, agents, projects, skills, MCP servers, and connectors—surfacing governance risks, producing audit reports, and deciding which agents to onboard into an identity-aware control plane with scoped permissions, just-in-time access, and full auditability. Learn more about **[aizome's integration with the Claude Compliance API](https://www.aizome.ai/integrations).** - **Akeyless Security:** Akeyless gives Claude agents secretless, just-in-time access to the systems they act on—databases, cloud services, and SaaS APIs—removing standing credentials and ensuring each agent performs only its permitted task, with a full audit trail of every action. To set up this integration, see the **[Akeyless Security setup guide](https://www.akeyless.io/blog/securing-ai-agents-in-claude-demo-and-setup-guide/)**. - **Akto:** Akto's Claude Compliance API integration gives security teams visibility into Claude Enterprise chats, files, connectors, and projects across AI agents, MCP servers, and human users—detecting prompt injection, sensitive-data exposure, and policy violations. To set up this integration, see the **[Akto setup guide](https://ai-security-docs.akto.io/akto-atlas-agentic-ai-security-for-employee-endpoints/endpoints-discovery-agents/anthropic-connector)**. - **Alignmt AI:** Alignmt AI transforms Compliance API activity into a risk-scored inventory of every AI resource, surfacing autonomous automations, write-capable connectors, and sensitive-data exposure, with provenance on every fact and human attestation for what evidence can't resolve. Learn more about **[Alignmt AI's integration with the Claude Compliance API](https://www.alignmt.ai/post/alignmt-ai-integrates-with-anthropic-s-compliance-api)**. - **Artemis Security:** Artemis ingests Claude Compliance API data to detect and investigate malicious behavior, correlating Claude activity with the rest of the environment for machine-speed detection and response. To set up this integration, see the **[Artemis setup guide](https://artemissecurity.com/company-news/secure-your-enterprise-ai-artemis-now-integrates-with-anthropic-compliance-api-and-telemetry/)**. - **Axonius:** Axonius brings Claude Enterprise into the Axonius Asset Cloud, reconciling directory and access artifacts against every other asset and exposure in your stack. Decision-grade asset intelligence surfaces coverage gaps, drift, and identity risk across the environment, mapped to the assets and owners behind them. To set up this integration, see the **[Axonius setup guide](https://docs.axonius.com/docs/anthropic)**. - **Bay:** Bay secures AI agents in the enterprise—providing real-time visibility and enforcement for Claude across all endpoints, containers, and browsers. To set up this integration, see the **[Bay setup guide](https://bay.security/use-cases/platform/claude).** - **Beacon Security:** Beacon is the platform for AI-native SecOps, serving as the data and context layer for security teams. Its Claude connector ingests Claude Enterprise activity and compliance data, normalizes and enriches it, and makes it available to security agents, SIEMs, data lakes, and compliance archives. Learn more about **[Beacon's integration with the Claude Compliance API](https://beacon.security/beacon-for-anthropic-compliance-api)**. - **BigID:** BigID secures Claude usage by giving security teams visibility into every file shared with Claude, pinpointing exactly which sensitive attributes are present the moment a risky file is uploaded. Learn more about **[BigID's integration with the Claude Compliance API](https://bigid.com/blog/securing-claude/)**. - **Bloom Security:** Bloom is the endpoint security platform for the AI-native era, giving security teams visibility and governance over every AI tool, plugin, and MCP server across their organization. The Claude Compliance API integration deepens that coverage—into every session, file, and configuration change across every user—so security teams can enable Claude at scale without losing control. To set up this integration, see the **[Bloom Security setup guide](https://bloom.security/anthropic-compliance-api-integration)**. - **Bold Security:** Prevent sensitive data from being misused by users or AI agents with real-time protection directly on the endpoint. To set up this integration, see the **[Bold Security setup guide](https://www.bold.security/blog/bold-security-extends-ai-data-protection-through-claudes-compliance-api)**. - **Brava Security:** Brava ingests the Claude activity feed via the Compliance API and turns it into fully-leveraged security telemetry—routed into your existing SIEM, data lake, and archive, and searchable for investigation—so one feed powers monitoring, insider risk, retention, and governance with no custom collectors. To set up this integration, see the **[Brava Security setup guide](https://brava.security/integrations/anthropic-claude)**. - **C1.ai:** C1.ai integrates with the Claude Compliance API to bring usage-informed access reviews, dormant-account detection, and automated lifecycle workflows to Claude Enterprise and Claude Platform—extending the same identity governance customers already run across the rest of their stack. To set up this integration, see the **[C1.ai setup guide](https://www.c1.ai/docs/baton/claude-enterprise)**. - **Capsule Security:** Capsule connects to the Claude Compliance API to turn your organization's members, conversations, and activity into an auditable AI inventory, giving security and compliance teams continuous visibility, data-leakage detection, rogue-agent detection, and audit-ready governance over Claude usage. To set up this integration, see the **[Capsule Security setup guide](https://docs.capsulesecurity.io/guides/claude-enterprise)**. - **Cato Networks:** Cato AI Security integrates with the Compliance API to secure, govern, and audit all Claude interactions, enforcing real-time policies, preventing data leakage, and maintaining comprehensive compliance visibility. To set up this integration, see the **[Cato Networks setup guide](https://www.catonetworks.com/integrations/ai-activity-monitoring/anthropic-compliance-api/)**. - **Check Point:** Check Point's Workforce AI Security integration with the Claude Compliance API gives enterprises audit-grade visibility into Claude usage across surfaces and organization-level MCP servers—per-user and per-surface usage analytics plus content-level exposure analysis across chats and uploaded files, flagging PII, credentials, and regulated data for existing security workflows. To set up this integration, see the **[Check Point setup guide](https://sc1.checkpoint.com/documents/Infinity_Portal/WebAdminGuides/EN/Workforce-AI-Security-Admin-Guide/Topics-Workforce-AI-Security-AG/Integrations.htm?tocpath=Integrations%7C_____0)**. - **Cloudflare:** Cloudflare’s CASB integration provides security teams agentless visibility into posture, data, and compliance risks across their organization’s use of Claude. To set up this integration, see the **[Cloudflare setup guide](https://developers.cloudflare.com/changelog/post/2026-05-19-casb-claude-compliance-api/)**. - **Clover Security:** Clover uses the Compliance API to give security teams organization-wide visibility into Claude Code usage—who's building with Claude Code, how much is already protected by Clover, and where the gaps are—then extends Clover's security controls to every developer. Learn more about **[Clover Security's integration with the Claude Compliance API](https://clover.security/blog/see-what-your-coding-agents-build-clover-integrates-with-anthropics-claude-compliance-api-to-give-security-real-time-visibility/)**. - **Concentric AI:** Concentric AI Semantic DLP gives organizations visibility into the AI tools employees use, including shadow AI. Through its integration with the Compliance API, it captures every prompt, response, and attachment—including historical activity across Claude's browser and desktop apps—to strengthen governance, accelerate investigations, and demonstrate compliance. Learn more about **[Concentric AI's integration with the Claude Compliance API](https://concentric.ai/integrations/anthropic-compliance/)**. - **Cribl:** The Claude Compliance API lets you ingest audit activity from Claude Enterprise and Claude Platform into Cribl Stream over HTTPS as a first‑class telemetry stream. It pulls user activities including authentication, chat interactions, file uploads, and administrative actions so you can normalize, enrich, mask, and route that data alongside the rest of your observability and security pipelines. To set up this integration, see the **[Cribl setup guide](https://docs.cribl.io/stream/sources-anthropic-compliance/)**. - **CrowdStrike:** CrowdStrike built an integration with the Claude Compliance API that brings Claude activity into the Falcon platform, delivering real-time visibility, detection, and automated response through Falcon Next-Gen SIEM and CrowdStrike’s Agentic SOAR. It will further extend protection with Falcon AIDR and Falcon Shield to continuously detect, govern, and secure AI usage at scale. To set up this integration, see the **[CrowdStrike setup guide](https://marketplace.crowdstrike.com/listings/anthropic-data-connector/)**. - **Cyberhaven:** Cyberhaven integrates with the Compliance API to give enterprise security teams visibility into the data their employees share with Claude, applying Cyberhaven's data classification engine to chats, files, and projects so customers can adopt Claude confidently while meeting their security and compliance requirements. To set up this integration, see the **[Cyberhaven setup guide](https://www.cyberhaven.com/support/anthropic-claude-enterprise-connector)**. - **Cyera:** Cyera integrates with Claude Enterprise through the Claude Compliance API, giving security teams comprehensive visibility into conversations, files, and user activity across the platform. Every interaction is analyzed through OmniDLP and enriched with Cyera’s 95%-precision data classification and risk scoring, helping teams identify sensitive data exposure, detect risky or unsanctioned behavior, and maintain audit-ready oversight from day one. Deployment takes just minutes. **[Learn more about the integration](https://www.cyera.com/blog/cyera-and-anthropic-integrate-to-bring-real-time-ai-security-to-claude-enterprise)** and contact your Cyera Sales Rep to get started. - **Dash Security:** Dash Security provides real-time, intent-aware monitoring and policy enforcement over Claude Enterprise conversations, detecting sensitive-data exposure, prompt injection, and risky behavior. By extracting conversational intent, Dash understands what users are actually doing with AI and distinguishes real threats from false positives and benign activity. To set up this integration, see the **[Dash Security setup guide](https://dash.security/integrations/claude-enterprise)**. - **Datadog:** Datadog’s Claude Compliance ingests audit logs from the Claude Platform into Datadog, giving security teams visibility into admin activity, API key lifecycle events, and authentication events across their Claude organization for SIEM and compliance use cases. To set up this integration, see the **[Datadog setup guide](https://docs.datadoghq.com/integrations/anthropic-compliance-logs/).** - **Daylight:** Monitor Claude usage in an organization for security investigations and compliance purposes. To set up this integration, see the **[Daylight setup guide](https://daylight.ai/claude-enterprise-detection)**. - **Elastic:** Collect Claude audit and usage activity to detect security risks, support compliance requirements, and investigate AI-related incidents. To set up this integration, see the **[Elastic setup guide](https://www.elastic.co/docs/reference/integrations/anthropic)**. - **Eon:** Eon provides AI governance for Claude through automated data classification and natural-language security queries, helping enterprises maintain visibility and control over sensitive data shared with AI tools. Learn more about **[Eon's integration with the Claude Compliance API](https://www.eon.io/blog/may-2026-eon-product-update)**. - **eSentire:** eSentire Atlas integrates with the Compliance API to deliver 24/7 managed detection and response for Claude usage. eSentire ingests activity events and conversation content, correlates them with endpoint, identity, and cloud telemetry, and investigates and responds to threats and policy violations on customers' behalf. Learn more about **[eSentire's integration with the Claude Compliance API](https://www.esentire.com/what-we-do/mdr-and-platform-integrations)**. - **Exaforce:** Exaforce ingests Claude Compliance API activity, chat, and file data into the Exaforce Agentic SOC Platform, giving security teams a governed, SOC-ready view of Claude usage to monitor, detect, and investigate risky activity alongside their broader security signals. To set up this integration, see the **[Exaforce setup guide](https://www.exaforce.com/blogs/anthropic-integration)**. - **ForceAI Security:** ForceAI PineSMITH integrates with the Anthropic Claude Compliance API to provide enterprise visibility into Claude activity, users, chats, files, projects, governance events, and audit-ready compliance data. To set up this integration, see the **[ForceAI Security setup guide](https://forceaisecurity.com/documents/claude-compliance-integration)**. - **Forcepoint:** Forcepoint's solution to securely enable AI now extends to Claude Enterprise via the Claude Compliance API, ingesting every conversation, file upload, generated content and user activity into a unified AI governance dashboard. Security teams can apply existing DLP and DSPM policies across all Claude interactions, detecting sensitive data exposures, enforcing compliance, and maintaining a complete audit trail, with historical activity backfilled the moment the connector is deployed. To set up this integration, see the **[Forcepoint setup guide](https://www.forcepoint.com/use-case/securely-enable-ai)**. - **Fortinet:** The Fortinet data security posture management integration with the Claude Compliance API enables organizations to monitor and analyze user interactions within Claude Enterprise, including prompts and uploaded files, to detect potential exposure of sensitive data such as PII, PHI, CHD, and authentication data. The integration provides visibility, alerting, and auditability for AI usage, helping security teams strengthen governance, compliance, and data protection across enterprise AI workflows. To set up this integration, see the **[Fortinet setup guide](https://www.fortinet.com/content/dam/fortinet/assets/alliances/fortidata-claude-setup-guide.pdf)**. - **Geordie AI:** Geordie, the agentic security and governance platform, extends its existing coverage of the Claude suite, including Claude Code and Claude Cowork, across the full Claude Enterprise deployment surface through the Claude Compliance API. Together, this gives security and compliance teams a single, coherent view of every Claude use case and activity across the enterprise, with the visibility and controls to govern it. To set up this integration, see the **[Geordie AI setup guide](https://www.geordie.ai/resources/geordie-extends-agentic-ai-security-and-governance-coverage-through-claude-compliance-api/)**. - **Glow:** Glow ingests the Compliance API activity feed to complete its picture of AI adoption and security posture across an organization's Claude workspace, enabling consistent governance and security controls across every form of Claude adoption, from individual users to managed deployments and agentic integrations. To set up this integration, see the **[Glow setup guide](https://docs.glow.io/integrations/anthropic)**. - **Grip Security:** Grip integrates with the Anthropic Compliance API to deliver SSPM checks, remediation, and AI agent visibility for Claude. To set up this integration, see the **[Grip Security setup guide](https://help.grip.security/docs/anthropic-claude-compliance-api-integration-with-grip-security-1)**. - **Hadrius:** Hadrius is the agentic compliance infrastructure for SEC- and FINRA-regulated firms. Its Claude integration captures conversations and activity into a compliant books-and-records archive with automated surveillance, so advisers and broker-dealers can adopt Claude while meeting their recordkeeping and supervision obligations. To set up this integration, see the **[Hadrius setup guide](https://support.hadrius.com/articles/8197408781-claude-integration?lang=en)**. - **Harmonic Security:** Harmonic Security uses the Anthropic Compliance API to govern Claude use across every surface (Desktop, Web, Mobile, Claude Code, Cowork), classifying every interaction with proprietary detection models to surface sensitive-data exposure, prompt injection, and AI usage and adoption insights broken down by team, department, and location. To set up this integration, see the **[Harmonic Security setup guide](https://docs.harmonicsecurity.app/integration-guides/configure-anthropic-compliance-api-connector)**. - **IBM Guardium (coming soon):** IBM Guardium helps organizations monitor data access and usage by agentic AI systems and help ensure autonomous activities remain within approved security boundaries. With ready-to-use templates, customizable policies, continuous audit workflows, and compliance reporting aligned to frameworks such as EU AI Act, AIDA, PCI, SOX, HIPAA, GDPR, and CPRA, Guardium helps security and compliance teams confidently demonstrate oversight of AI-driven activity. Learn more about **[IBM Guardium](https://www.ibm.com/products/guardium)**. - **Island:** Island's Claude Compliance API integration gives security and compliance teams visibility into every Claude session—user-driven or autonomous—covering conversation topics, tool/model/skill usage, and continuously enriched detection of sensitive-data exposure and policy violations for posture assessment and anomaly detection. Learn more about **[Island's integration with the Claude Compliance API](https://www.island.io/blog/island-integrates-with-the-claude-compliance-api)**. - **Linx Security:** Linx integrates with the Claude Compliance API to govern access across human, non-human, and agent identities in Claude—delivering visibility, risk detection, access reviews, and least-privilege enforcement. To set up this integration, see the **[Linx Security setup guide](https://docs.linxsecurity.io/connectors/connectors-library/anthropic/claude.ai)**. - **Microsoft Purview:** Purview's integration with the Claude Compliance API extends Purview's enterprise security and compliance coverage by surfacing insights from Claude activity feed and chat conversations, enabling scenarios within Purview Data Security Posture Management and Audit. It provides centralized visibility of activities across Claude Enterprise and Claude Platform, allowing organizations to understand how data is accessed and used across their Claude systems. Surfacing this signal within Purview helps close a critical governance gap as enterprises rapidly adopt AI across multiple platforms. The same comprehensive visibility Purview provides for Copilot and the broader Microsoft AI ecosystem, is now available for Claude Enterprise users. To set up this integration, see the **[Microsoft Purview setup guide](https://learn.microsoft.com/en-us/purview/ai-microsoft-purview)**. - **Mimecast:** Mimecast's Governance, Compliance & Insights (GCI) platform now integrates with the Claude Compliance API to bring Claude Enterprise conversations into the same unified governance and compliance experience organizations already use for email and collaboration data, with no new interfaces or workflows required. To set up this integration, see the **[Mimecast setup guide](https://www.mimecast.com/use-cases/ai-governance/)**. - **Mint Security:** Mint provides contextual runtime security for AI agents, using the Claude Compliance API to apply organization-specific visibility, context, and policy controls to how teams use Claude. To set up this integration, see the **[Mint Security setup guide](https://docs.mint.security/integrations/claude/claude)**. - **Netskope:** Netskope's integration with the Claude Compliance API lets organizations connect Claude Enterprise directly into the enhanced AI security features of the Netskope One platform, delivering full asset, identity, and activity visibility, robust policy enforcement, advanced data security, and comprehensive security posture management, all within the workflows security and compliance teams already rely on. **Note:** Available in private preview. To set up this integration, see the **[Netskope setup guide](https://docs.netskope.com/en/onboard-claude-for-ai-discovery/)**. - **Nightfall AI:** Nightfall integrates with the Claude Compliance API to detect sensitive data (PII, PHI, PCI, secrets, source code) in Claude Enterprise chats and file uploads. Monitor-only, with admin alerts, end-user notifications, and an audit trail in the same workflow as other SaaS connectors. To set up this integration, see the **[Nightfall AI setup guide](https://help.nightfall.ai/nightfall-for-anthropic-claude/nightfall-for-claude-enterprise-compliance-api)**. - **Noma Security:** Noma's AI Detection and Response (AI-DR) integration with the Compliance API gives security teams continuous visibility into Claude Enterprise usage, detecting sensitive-data exposure, risky prompts, and policy violations with full behavioral context—so teams can say yes to Claude with governance in place. Learn more about **[Noma Security's integration with the Claude Compliance API](https://noma.security/integrations/claude-enterprise/)**. - **Obsidian Security:** Obsidian extracts posture and activity from the Compliance API and 200+ other applications. That data is normalized and used by enterprise customers for continuous governance, drift detection, and identifying and investigating potential threats. Learn more about **[Obsidian Security's integration with the Claude Compliance API](https://www.obsidiansecurity.com/integrations/claude)**. - **Okta:** Okta's integration with the Claude Compliance API brings Claude usage within Okta Identity Security Posture Management, allowing security and compliance teams to contextualize Claude user activity with signals from other identity platforms, SaaS tools, and cloud providers and surface identity blind spots from one unified view. This enables organizations to continuously assess identity risk posture, detect misconfigurations and risks—such as overprivileged users, dormant accounts, and unused admin keys—remediate issues by severity, and securely deploy AI at scale. Learn more about **[Okta's integration with the Claude Compliance API](https://www.okta.com/newsroom/articles/okta-identity-security-posture-management-integrates-with-anthropics-new-compliance-api/)**. - **Note on availability:** This offering will soon be available in beta to select customers using Okta Identity Security Posture Management alongside either Anthropic’s Claude Enterprise or Claude Platform. - **Onyx Security:** Onyx is the secure AI control plane that gives enterprises observability, security, and governance over AI agents across SaaS, cloud, endpoint, and code. This integration extends Onyx's guardrails and audit trail to Claude deployments—letting security teams monitor usage, enforce policies, prevent data exfiltration, and meet compliance requirements without slowing Claude adoption. To set up this integration, see the **[Onyx Security setup guide](https://onyx.security/integrations/anthropic)**. - **Opsin:** AI risk and agent governance for Claude Enterprise: Opsin's Compliance API integration surfaces data oversharing, risky user behavior, and agent sprawl within 24 hours, using a context graph that connects identity, data, and model for root-cause remediation. To set up this integration, see the **[Opsin setup guide](https://www.opsinsecurity.com/claude-integration)**. - **Orca Security:** Orca Security extends its unified platform to Claude Enterprise, giving security teams complete visibility into Claude identities, access, AI workspaces, and the sensitive data within them—to surface risk, enforce least privilege, and track compliance. To set up this integration, see the **[Orca Security setup guide](https://claude-doc.orca.security/)**. - **Origin:** Origin is endpoint observability for agentic AI: every AI agent on your fleet, what it did, and where your AI spend goes. The Compliance API adds Claude Enterprise usage from claude.ai and Claude Desktop, attributed to each user and folded into the same record as the agents Origin observes. To set up this integration, see the **[Origin setup guide](https://support.originhq.com/docs/claude-enterprise)**. - **Palo Alto Networks:** The Claude Compliance API integration enables organizations to understand what data is being shared with Claude, detect sensitive content and policy violations, identify threats and anomalous behavior, and guide users toward safer AI usage. To set up this integration, see the **[Palo Alto Networks setup guide](https://www.paloaltonetworks.com/blog/cloud-security/claude-security-integration-ai-governance)**. - **Pluto Security:** Pluto provides real-time visibility and enforcement across the Anthropic ecosystem—Claude Code, Cowork, and Office Agents. It monitors agent activity, detects risks like prompt injection and sensitive-data exposure, identifies malicious ecosystem components, and enforces policies on agent actions, MCPs, connectors, and tool usage. Learn more about **[Pluto Security's integration with the Claude Compliance API](https://pluto.security/blog/claude-enterprise-meets-ai-security-platform/)**. - **Proofpoint:** Proofpoint’s integration with the Claude Compliance API brings security and governance together across AI interactions. Sensitive data in prompts, responses, and files is protected with DLP, AI-aware classifiers, and Insider Threat signals, while activity is captured and retained for eDiscovery, supervision, and compliance purposes. The result is comprehensive visibility, control, and retention for organizations leveraging Claude, consistent with the controls protecting the rest of their tech stack. To set up this integration, see the **[Proofpoint setup guide](https://support.nuclei.ai/hc/en-us/articles/39717913412759-Compliance-Archiving-for-Anthropic-Claude-Enterprise)**. - **Reco:** Reco monitors Claude Enterprise for AI security threats, posture risks, and identity anomalies. Using the Compliance API, it ingests activity logs and conversation data to detect MCP server abuse, permission escalations, and unauthorized access, surfacing violations mapped to MITRE ATT&CK, OWASP LLM Top 10, and more. Learn more about **[Reco's integration with the Claude Compliance API](https://www.reco.ai/blog/claude-security-governance)**. - **RelativityOne:** RelativityOne enables legal, compliance, and security teams to collect Claude Enterprise data directly from the source via the Claude Compliance API. This includes conversation prompt and response data—such as chats, uploaded files, and projects. Collected data is automatically mapped, preserved, and processed into a reviewable format, including normalization into Relativity Short Message Format (RSMF). This ensures teams can quickly analyze conversational evidence with full context while maintaining the metadata and transparency required for defensibility. To set up this integration, see the **[RelativityOne setup guide](https://help.relativity.com/RelativityOne/Content/Collect/Data_sources/Claude.htm)**. - **ReliaQuest:** The ReliaQuest GreyMatter platform now integrates with the Claude Compliance API, bringing Anthropic’s Claude Enterprise activity data into GreyMatter, giving security and IT teams visibility into activity logs to detect threats or unwanted behavior. To set up this integration, see the **[ReliaQuest setup guide](https://greymatter.myreliaquest.com/resources/articles/anthropic-claude-compliance)**. - **Rubrik:** Rubrik Agent Cloud integrates with the Claude Compliance API to help organizations control how Claude is used at work. Teams can see who is using Claude, review activity when needed, flag risky behavior, and use Rubrik SAGE to create natural-language policies that route alerts to their security tools. To set up this integration, see the **[Rubrik setup guide](https://www.rubrik.com/products/rubrik-agent-cloud)**. - **Safeguard:** Safeguard integrates with the Claude Compliance API to monitor Claude usage as part of AI-native software supply chain security—surfacing vulnerabilities and risks with AI-driven remediation across the lifecycle. To set up this integration, see the **[Safeguard setup guide](https://safeguard.sh/mcp/claude)**. - **SailPoint:** The SailPoint Claude Compliance API integration streamlines identity and access management by aggregating Organizational users, groups, and roles. To set up this integration, see the **[SailPoint setup guide](https://documentation.sailpoint.com/connectors/saas/claude_enterprise/help/)**. - **Saviynt:** The Saviynt Claude Compliance API integration streamlines identity governance for Claude Enterprise by aggregating users, groups, and roles to enable centralized visibility, automated access certifications, and seamless lifecycle management. To set up this integration, see the **[Saviynt setup guide](https://exchange.saviynt.com/products/claude-enterprise)**. - **SentinelOne:** Give security teams visibility and control to adopt Claude across the enterprise with SentinelOne's Anthropic Compliance API integration. Prompt Security scans every prompt and response against your policies, flagging violations in real time. Singularity AI SIEM correlates Claude activity across your security telemetry to accelerate investigation, detection, and response. Learn more about **[SentinelOne's integration with the Claude Compliance API](https://sentinelone.com/blog/sentinelone-and-claude-integrations-for-ai-visibility-governance-and-defense)**. - **Sentra:** Discover sensitive data across projects, chats, prompts, and files, identify data exposure risks, and extend data security and compliance controls to your AI environments. To set up this integration, see the **[Sentra setup guide](https://sentra.io/guides/sentra-claude-enterprise-integration-guide)**. - **Smarsh:** Smarsh Capture helps organizations monitor and retain Claude Enterprise interactions, so you can adopt and govern AI with confidence and support compliance workflows. By connecting directly to the Claude Compliance API, you gain structured access to Claude conversations, consistent complete records, and support for audit and regulatory response. This enables company-wide Claude Enterprise adoption and governance while staying ahead of evolving regulations. To set up this integration, see the **[Smarsh setup guide](https://central.smarsh.com/s/article/From-which-networks-does-cloudcapture-gather-data-from)**. - **Snyk:** This new integration gives security and compliance teams a complete inventory of their Claude Enterprise environment including models, approved MCP servers, and tool-level permissions, in the platform they already use to govern the rest of the stack. To set up this integration, see the **[Snyk setup guide](https://docs.snyk.io/integrations/cloud-ai-platforms/anthropic-claude-enterprise)**. - **Sola Security:** Sola connects to the Compliance API to bring Claude Enterprise activity, user access, and conversation data into Sola's AI-powered security platform, where security teams can query their Claude environment in plain language, correlate it with identity tools, and surface governance risks like overprivileged users and orphaned accounts. To set up this integration, see the **[Sola Security setup guide](https://docs.sola.security/integrations/data-sources/claude-enterprise)**. - **Strac:** Strac helps enterprises safely adopt Claude by using the Compliance API to detect sensitive-data exposure, monitor usage, and generate audit-ready security insights. To set up this integration, see the **[Strac setup guide](https://www.strac.io/integrations/claude-compliance-api)**. - **Straiker:** Straiker provides real-time security for Claude Enterprise—detecting prompt injection, data exfiltration, and agentic misuse across Claude chat and Claude Code. Security teams get unified visibility, enforcement, and forensic investigation across all Claude surfaces in a single platform. Learn more about **[Straiker's integration with the Claude Compliance API](https://www.straiker.ai/claude-code)**. - **Sumo Logic:** Sumo Logic’s integration with the Claude Compliance API brings Claude Enterprise activity logs and Claude platform activity logs into the Sumo Logic Intelligence Operations Platform, so teams can transform AI governance into a transparent, enterprise-ready operation that reduces risk through real-time monitoring and automated detection, investigation, and response capabilities within Sumo Logic. To set up this integration, see the **[Sumo Logic setup guide](https://www.sumologic.com/help/docs/integrations/saas-cloud/claude-compliance)**. - **Surf AI:** Surf reads Claude usage from the Compliance API to detect unsanctioned MCP connections, tie each to an accountable owner, and drive remediation. Learn more about **[Surf AI's integration with the Claude Compliance API](https://www.surf.ai/technology-partners/anthropic)**. - **Sysdig:** Sysdig detects suspicious Claude credential activity and correlates it with runtime and security events on the same machine, turning isolated API alerts into clear evidence of misuse and allowing security teams to respond to real threats with full context. To set up this integration, see the **[Sysdig setup guide](https://www.sysdig.com/integrations/anthropic)**. - **Tenable:** The integration with the Claude Compliance API will leverage visibility into user interactions to detect malicious and suspicious activity. This capability allows organizations to safely adopt Claude Enterprise across the organization without losing visibility into how users engage with it. To set up this integration, see the **[Tenable setup guide](https://docs.tenable.com/early-access/ai-exposure/Content/settings/integrations/claude.htm)**. - **Theta Lake:** As enterprises deploy Claude across their workforce to unlock productivity and ROI at scale, AI interactions between AI and humans and AI agents are causing entirely new user behaviors and generating a new class of communications. Security, compliance, and legal teams need complete visibility into how Claude is being used. This integration empowers organizations to use Theta Lake’s Investigation, Analytics, and Regulatory Compliance capabilities, each of which received the tops scores in the Gartner® Critical Capabilities. To set up this integration, see the **[Theta Lake setup guide](https://docs.thetalake.com/docs/en-US/integrations/anthropic.html?Highlight=anthropic%20)**. - **Token Security:** Token Security integrates with Anthropic's Compliance API to discover, govern, and secure the non-human identities powering Claude usage, surfacing stale keys, ownership gaps, and agent activity within a unified NHI security platform. Learn more about **[Token Security's integration with the Claude Compliance API](https://www.token.security/blog/token-security-brings-anthropic-claude-compliance-api-into-its-identity-control-plane)**. - **Torch Security:** Torch brings Claude Enterprise under identity governance: it reads the Claude org directory and activity feed via the Compliance API, correlates every account to a real identity, and runs access reviews that flag out-of-band (non-SCIM) group grants, orphaned accounts, and dormant access—read-only, never touching conversation content. To set up this integration, see the **[Torch Security setup guide](https://www.torch.security/integrations/claude/)**. - **Trellix:** The Trellix Claude Compliance API integration checks all chat content, artifacts, and responses against customer-defined privacy policies using their Data Security Engine and creates detection and audit events in Trellix Helix. This includes customer-specific patterns and policies that are defined for their endpoint, server, and collaboration software and applies it to Claude as well. To set up this integration, see the **[Trellix setup guide](https://www.trellix.com/platform/data-security/)**. - **TrendAI:** Trend Micro's integration with the Claude Compliance API automatically collects and monitors Claude chat activity across your organization, scans for prompt injection, harmful content, and sensitive-data leakage, and correlates findings with your broader security signals through SIEM and XDR for investigation and audit-ready compliance and governance. To set up this integration, see the **[TrendAI setup guide](https://www.trendmicro.com/en_us/research/26/f/governing-claude-enterprise.html)**. - **Valence Security:** Valence integrates with the Compliance API to give security teams visibility into how Claude Enterprise and Claude Platform are used, including users, agents, MCP servers, connectors, and tools. Valence correlates this activity with identities and configurations across the rest of the environment, helping teams detect risky behavior and remediate exposure. To set up this integration, see the **[Valence Security setup guide](https://www.valencesecurity.com/resources/blogs/securing-agentic-identities-in-claude-the-valence-integration-with-the-compliance-api)**. - **Varonis:** The Compliance API integration deepens Varonis' support for Claude, enabling security and governance teams to monitor usage, investigate misuse across full sessions, and assess AI-related risk with data context. To set up this integration, see the **[Varonis setup guide](https://info.varonis.com/hubfs/documentation/varonis-atlas-anthropic-compliance-api-documentation.pdf)**. - **Vega:** Vega ingests your organization's Compliance API activity feed, normalizes it to a common security schema, and lets your team search, detect, alert, and investigate Claude audit events alongside the rest of their security telemetry. Learn more about **[Vega's integration with the Claude Compliance API](https://vega.io/blog/vega-anthropic-claude-compliance-api)**. - **Vero Security:** Vero Security integrates with the Compliance API to provide security insights for enterprises, enabling full context for governance and policy enforcement so teams can monitor, manage, and safely scale Claude adoption across the organization. Learn more about **[Vero Security's integration with the Claude Compliance API](https://www.vero.security/blog/vero-antrpc-integration)**. - **Willow:** Willow gives every AI agent an identity and permissioned access to the right tools and skills, with a full audit trail. Through its integration with the Compliance API, Willow extends that governance to Claude usage so security teams can see and control agent activity across the organization. To set up this integration, see the **[Willow setup guide](https://docs.withwillow.ai/docs/admin/settings/claude-compliance-sync)**. - **Wiz:** The Wiz connector for the Claude Compliance API allows organizations to discover AI projects, models, and users while identifying security risks like elevated roles or sensitive data within knowledge bases. It maps identity relationships and tracks model usage across the enterprise by integrating via the Claude Compliance API. Learn more about **[Wiz's integration with the Claude Compliance API](https://www.wiz.io/blog/claude-wiz-integration)**. - **Zenity:** Zenity automatically discovers and inventories all Claude agents, MCP servers, connected resources, and non-human identities—mapping their relationships and security posture from the moment a session starts. To set up this integration, see the **[Zenity setup guide](https://zenity.io/use-cases/platform/claude-enterprise)**. - **Zscaler:** By integrating with the Claude Compliance API, Zscaler is now able to provide deep visibility and granular control over every Claude prompt created by an enterprise user. Zscaler Data Security and DLP workflows help administrators strengthen their data protection programs through detailed analysis of organizational data and activity logs extracted from Claude. This solution enables faster enforcement actions to prevent data loss across multiple exfiltration channels. To set up this integration, see the **[Zscaler setup guide](https://www.zscaler.com/products-and-solutions/data-security)**. --- ## get-started-with-claude-in-slack-ad4d7e43e7.md # Get started with Claude in Slack Claude in Slack will be switched over to the new Claude Tag experience on August 3, 2026. To integrate Claude and Slack, use Claude Tag instead. Learn more **[about Claude Tag](https://claude.com/product/tag)**. You can now integrate Claude and Slack, giving you two ways to use them together: add Claude directly to your Slack workspace, or enable the Slack connector for your Claude apps. ## What is Claude in Slack? The Claude app is available to users on paid Slack plans. Slack admins must approve the Claude app before individual users can access it. It’s how we’ve brought Claude’s capabilities directly to Slack, bringing AI assistance into your team’s workspace. This integration allows you to work with Claude without leaving Slack through three convenient surfaces: **Direct message with Claude**: Start a private conversation with @Claude. **AI assistant panel**: Click the Claude icon in Slack's AI assistant header to open a panel on the right side of your Slack window, allowing you to access Claude from anywhere in the Slack app. **Thread participation**: Mention @Claude in any thread to get Claude's help with the conversation. All surfaces provide the same capabilities that you have enabled in Claude, including web search and connections to your integrated tools, allowing you to seamlessly integrate AI assistance into your existing workflow. **Note:** Team and Enterprise plan users with access to Claude Code on the web can also route coding tasks directly to Claude Code by mentioning @Claude. See [Using Claude in Slack](https://support.claude.com/en/articles/12461605-using-claude-in-slack#h_adda66b697) for details on this beta feature. ------------------------------------------------------------------------ ## Enable and install Claude in Slack ### For Slack admins 1. Go to the [Claude app in the Slack App Marketplace](https://slack.com/marketplace/A08SF47R6P4). 2. Click "Add to Slack" on the Claude app page. 3. Review and approve the app for your organization. 4. Choose whether to deploy org-wide or to specific workspaces. **To install across all workspaces:** 1. Navigate to your Slack management workspace at: [https://app.slack.com/manage/\/workspaces/all](https://app.slack.com/manage/%3CINSERT_SLACK_ID%3E/workspaces/all) - Find your enterprise's Slack ID using the appropriate lookup method 2. Navigate to **Integrations → Installed apps → Add to more workspaces**. 3. Toggle through all relevant workspaces where you'd like to enable Claude. ### For individual users Once your Slack admin has approved Claude (or if you're on a personal Slack plan): 1. Find Claude in your apps list (search for "Claude" if it's not immediately visible), or go to the Slack App Marketplace. 2. Click "Connect Account” to be prompted to connect your Claude account: 3. In the window that opens, select which organization you would like to connect with Claude for Slack. 4. Click “Authorize” to allow Claude in Slack to access your Claude chat account: 5. You should see a confirmation message upon successful connection: 6. After successful authentication, return to Slack. 7. Click “+ New Chat” to start a conversation with Claude, or @mention Claude in any Slack conversation to access its capabilities. **Tip**: Add Claude to your Slack header for quick access by clicking the three dots "..." at the top right and selecting "Add this app to header." **Enable Claude Code in Slack (beta)** Team and Enterprise plan users can route coding tasks from Slack to Claude Code on the web. To enable this capability: 1. A Claude Owner or Primary Owner must enable Claude Code on the web by navigating to **[Organization settings \> Claude Code](http://claude.ai/admin-settings/claude-code)**. 2. Individual users must have access to Claude Code on the web. Once enabled, mentioning @Claude for coding tasks will automatically create a Claude Code session. Learn more about **[using Claude Code in Slack](https://support.claude.com/en/articles/12461605-using-claude-in-slack#h_adda66b697)**. ------------------------------------------------------------------------ ## What is the Slack connector? The Slack connector is available for all paid plans (Pro, Max, Team, and Enterprise). Enabling the Slack connector allows Claude to search within your Slack workspace’s channels, direct messages, and shared files to pull relevant context into your conversations. Note that members of Team and Enterprise plan organizations will not see the option to enable the Slack connector individually until it’s enabled by an Owner. **Important:** You must install Claude in Slack before enabling and using the Slack connector. ## Enable the Slack connector ### Team and Enterprise owners 1. Log in to your Owner or Primary Owner account and click your initials in the lower left corner. 2. Navigate to **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. 3. Under “Connectors,” click the "Enable" button next to the Slack connector. 4. Users can then authenticate in their individual connector settings to begin using Slack in Claude. ### Individual Pro, Max, Team, and Enterprise users 1. Log in to your Claude account and click your initials in the lower left corner. 2. Navigate to **[Settings \> Connectors](http://claude.ai/settings/connectors)**. 3. Find the Slack connector and click “Connect.” 4. Click "Connect" to authenticate with the connector and start using Slack in Claude. ------------------------------------------------------------------------ ## Manage your Claude in Slack connections ### View Claude app connection status 1. Click on the Claude app in your Slack sidebar. 2. Go to the "Home" tab. 3. You'll see your connection status, including your connected account email and organization name. ### Disconnect the Claude app To disconnect your Claude account from Slack: 1. Go to the Claude Home tab in Slack. 2. Under **Disconnect Claude Account**, click the red "Disconnect" button. 3. Confirm the disconnection. Disconnecting will: - Remove the connection between your Claude account and Slack workspace. - Delete all past Claude conversations in Slack from Claude (within 30 days). - Preserve conversations in Slack, but Claude won't have awareness of them if you reconnect ### Disconnect the Slack connector You can also disconnect the Slack connector from your Claude settings (or you can enable/disable the connector for an individual chat): 1. Go to [](https://claude.ai/settings/integrations) claude.ai/settings/connectors 2. Find **Slack** in your list of Connectors. 3. Click the menu (...) and select "Disconnect." ------------------------------------------------------------------------ ## Privacy and data ### Data storage Your Slack conversations with Claude remain separate from your Claude history, keeping work organized across platforms. ### Data visibility - Conversations initiated in Slack are not visible in [your Claude chat history](http://claude.ai/recents). - Conversations initiated in the Claude web app are not accessible in Slack. - Each platform maintains separate conversation histories. ### Data deletion - Conversations are automatically deleted from Claude within 30 days if you disconnect the integration or uninstall the app. - Your conversations in Slack follow your organization's Slack retention policies. ------------------------------------------------------------------------ ## Frequently asked questions ### I’m trying to add Claude in Slack but it’s not working – help! If you are using a company Slack instance and are not assigned to an Admin role, a Slack Admin must approve the Claude app on behalf of your organization before you’re able to download it. If you try to skip this step and install Claude in Slack, you’ll see a **Request to install** prompt where you can send a message to your Slack Admin. Work with them to approve the app and make it available for your team. --- ## get-started-with-custom-connectors-using-remote-mcp-dcba133490.md # Get started with custom connectors using remote MCP Custom connectors using remote MCP are available on Claude, Cowork, and Claude Desktop for users on Free, Pro, Max, Team, and Enterprise plans. Free users are limited to one custom connector. ## What are custom connectors? Custom connectors let you connect Claude directly to the tools and data sources that matter most to your workflows. This enables Claude to operate within your favorite software and draw insights from the complete context of your external tools. You can: - Connect Claude to existing remote MCP servers. - Build your own remote MCP servers to connect with any tool. **Security and privacy with custom connectors** Custom connectors allow you to connect Claude to services that haven't been verified by Anthropic. Once connected, Claude can access those services and take action in them. For more guidance, review the **[Security and privacy considerations](#h_b79c05dfcd)** section below. ## What are remote MCP servers? The Model Context Protocol (MCP) is an open standard, created by Anthropic, for AI applications to connect to tools and data. Previously, **[MCP servers only ran locally](https://support.claude.com/en/articles/10949351-getting-started-with-local-mcp-servers-on-claude-desktop)** (i.e. on a user's laptop). Now, developers can build and host remote MCP servers that communicate with AI apps over the internet. Remote MCP servers give models access to internet-hosted tools and data, transforming Claude into an informed teammate that can independently handle complex, multi-step projects tailored to your needs. ------------------------------------------------------------------------ ## Network requirements When you add a custom connector, Claude connects to your remote MCP server from Anthropic's cloud infrastructure, rather than from your local device. This is true across every Claude client, including claude.ai, Claude Desktop, Cowork, and the mobile apps. This means your MCP server must be reachable over the public internet from Anthropic's IP ranges. Servers hosted on a private corporate network, behind a VPN, or blocked by a firewall won't connect, even if you can reach them from your own machine. ### If your server is on a private network You’ll need to allowlist Anthropic's IP addresses in your firewall so inbound connections from Claude can reach your server. See **[Anthropic IP addresses](https://platform.claude.com/docs/en/api/ip-addresses)** for the current ranges. ### Why this applies to Cowork and Claude Desktop Even though Cowork and Claude Desktop run on your computer, remote connectors are configured and brokered through your Claude account. The connection to your MCP server originates from Anthropic's servers, not from your machine's network interface. Local MCP servers configured in Claude Desktop via `claude_desktop_config.json` are a separate mechanism and do use your local network, but those aren't available in Cowork or claude.ai. ------------------------------------------------------------------------ ## Add a custom connector **Note:** While anyone can build and host connectors using remote MCP, only Owners can add them to Team and Enterprise plans. Once a connector has been added to a Team or Enterprise organization, users individually connect to and enable that connector. This ensures that Claude can only access tools and data that the individual user has access to. ### For Team and Enterprise plans **Preliminary steps for owners:** Before members of Team and Enterprise plans can configure custom connectors, an Owner or Primary Owner needs to follow these initial steps to add a custom connector to your organization: 1. Navigate to **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. 2. Click the "Add" button. 3. Hover over “Custom,” then select “Web.” 4. Add your connector's remote MCP server URL. 5. Optionally, click “Advanced settings” to specify an OAuth Client ID and OAuth Client Secret for your server. 6. Finish configuring your connector by clicking "Add." **Steps for members after connector is configured:** 1. Navigate to **[Customize \> Connectors](https://claude.ai/customize/connectors)** 2. Find the custom connector your Owner added in the list (it will have a "Custom" label). 3. Click "Connect" to authenticate and start using the connector with Claude. ### For Pro and Max plans If you are using an individual Pro or Max plan, follow these steps to add a custom connector: 1. Navigate to **[Customize \> Connectors](https://claude.ai/customize/connectors)** 2. Click "+" then “Add custom connector.” 3. Add your connector's remote MCP server URL. 4. Optionally, click “Advanced settings” to specify an OAuth Client ID and OAuth Client Secret for your server. 5. Finish configuring your connector by clicking "Add." ### Enabling connectors after configuration You can enable connectors for individual conversations via the “+” button on the lower left of your chat interface, then "Connectors." You'll see your configured connectors with toggles allowing you to enable/disable them per conversation. ------------------------------------------------------------------------ ## Remove custom connectors You can remove a custom connector by following these steps: 1. Navigate to **[Customize \> Connectors](https://claude.ai/customize/connectors)** 1. Team and Enterprise Owners can do this on their organization's behalf in **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)** 2. Locate the "Connectors" section. 3. Click "Remove" or select the three dots next to the connector you'd like to remove. 4. Follow the prompts to remove. If you're hoping to edit a custom connector, you'll need to remove it first, then re-add it using the updated details. ------------------------------------------------------------------------ ## Build custom connectors To learn about building connectors to use with Claude, see **[Building custom connectors via remote MCP servers](https://support.claude.com/en/articles/11503834-building-custom-connectors-via-remote-mcp-servers)**. ------------------------------------------------------------------------ ## Security and privacy considerations Custom connectors allow you to connect Claude to arbitrary services that have not been verified by Anthropic. When you connect Claude to external services, you're granting it the ability to access and potentially modify data within those services based on your permissions. It’s important to make sure you’re only connecting to remote MCP servers that you trust and that you’re aware of Claude’s interactions with web connectors. ### Security and permissions When you add a custom connector to Claude, you'll typically go through an OAuth authentication process to securely sign in to the application and grant specific permissions. This allows Claude to interact with the application on your behalf, without Claude ever seeing your actual password. You can revoke these permissions at any time by disconnecting the connector in Claude's settings or the third-party service's security settings. Remote MCP servers act as intermediaries between Claude and external applications. You should: - **Only connect to trusted servers:** Only connect Claude to servers built and hosted by organizations and applications you trust. - **Review requested permissions carefully:** During auth, review what permissions the MCP server is requesting to the application. Limit these scopes when possible and deny access if requested permissions seem unnecessary. - **Be aware of prompt injections:** Malicious MCP servers may include hidden instructions that try to make Claude perform unintended actions. Claude has built-in protections that attempt to block these attacks, but it's important to pay attention to tool inputs & outputs and connect only to trusted servers. - **Monitor changes in tool behavior:** Server developers may update tool behavior unexpectedly, leading to unintended or malicious behavior. ### Reporting malicious MCP servers If you become aware of a malicious MCP server, please report it to our **[vulnerability disclosure program](https://hackerone.com/anthropic-vdp/)**, and choose [`https://github.com/modelcontextprotocol`](https://github.com/modelcontextprotocol) as the Asset. ### Taking actions with tools Remote MCP servers give Claude tools it can invoke during your conversation. The developer of an MCP server can define what these tools do, including: - Reading data from connected applications. - Creating, modifying, or deleting data in connected applications. - Taking actions on behalf of the user. Claude can only access resources that you've given the server permission to access, but you should: - Be aware of any actions Claude is taking and that they have no destructive or unintended effects. - Review Claude's tool approval requests carefully and only click "Allow always" when using a server and tool that you trust to run unsupervised. - Using the "Search and tools" menu, disable any tools that aren't relevant to the current conversation or that you don't want Claude to be able to invoke. ### Interactive connectors Some connectors can display interactive interfaces directly within your Claude conversations. Instead of only returning text-based responses, these connectors can open live, interactive apps — like dashboards, task boards, or design tools — right in the chat. Interactive connectors appear in two ways: - **Inline cards:** Compact components embedded in the conversation, showing summaries, confirmations, or quick actions. - **Fullscreen view:** Immersive interfaces for complex interactions like data visualizations or document editing. The conversation composer remains available so you can continue chatting with Claude. You can interact with these connectors directly — filtering data, checking off tasks, adjusting settings — without leaving the conversation. Any actions you take within the interface use the same permissions you granted when connecting the tool. **Admin controls:** Team and Enterprise plan owners can disable specific tool calls that render interactive connectors within **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. ### Using Claude with Research **Note: [Advanced Research](https://claude.com/blog/integrations)** is not currently able to invoke tools from local MCP servers. Research allows Claude to deeply investigate queries by searching through hundreds of internal and external sources. During the research process, Claude can invoke tools from your connectors automatically without further approval. When using Research with custom connectors: - Disable any tools that can take write actions in external applications. - Review Claude’s approval request carefully and be aware of which tools you’re granting Claude permission to invoke. - Be mindful of the impact of Claude sending a large number of requests to your connectors. See **[Using Research on Claude](https://support.claude.com/en/articles/11088861-using-research-on-claude)** for more information about this feature. --- ## get-started-with-custom-connectors-using-remote-mcp.md # Get started with custom connectors using remote MCP Updated today Custom connectors using remote MCP are available on Claude, Cowork, and Claude Desktop for users on free, Pro, Max, Team, and Enterprise plans. Free users are limited to one custom connector. This feature is currently in beta. ## What are custom connectors? Custom connectors let you connect Claude directly to the tools and data sources that matter most to your workflows. This enables Claude to operate within your favorite software and draw insights from the complete context of your external tools. You can: - Connect Claude to existing remote MCP servers. - Build your own remote MCP servers to connect with any tool. **⚠️ Security and Privacy with Custom Connectors (beta)** Be aware that custom connectors allow you to connect Claude to services that have not been verified by Anthropic, and allow Claude to access and take action in these services. For more guidance, review the **[Security and privacy considerations](#h_9088ccdf4d)** section below. ## What are remote MCP servers? The Model Context Protocol (MCP) is an open standard, created by Anthropic, for AI applications to connect to tools and data. Previously, **[MCP servers only ran locally](https://support.claude.com/en/articles/10949351-getting-started-with-local-mcp-servers-on-claude-desktop)** (i.e. on a user's laptop). Now, developers can build and host remote MCP servers that communicate with AI apps over the internet. Remote MCP servers give models access to internet-hosted tools and data, transforming Claude into an informed teammate that can independently handle complex, multi-step projects tailored to your needs. ## Add a custom connector **Note:** While anyone can build and host connectors using remote MCP, only Owners can add them to Team and Enterprise plans. Once a connector has been added to a Team or Enterprise organization, users individually connect to and enable that connector. This ensures that Claude can only access tools and data that the individual user has access to. ### For Team and Enterprise plans **Preliminary steps for owners:** Before members of Team and Enterprise plans can configure custom connectors, an Owner or Primary Owner needs to follow these initial steps to add a custom connector to your organization: 1. Navigate to **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. 2. Click "Add custom connector" at the bottom of the section. 3. Add your connector's remote MCP server URL. 4. Optionally, click “Advanced settings” to specify an OAuth Client ID and OAuth Client Secret for your server. 5. Finish configuring your connector by clicking "Add." **Steps for members after connector is configured:** 1. Navigate to **[Settings \> Connectors](https://claude.ai/settings/connectors)**. 2. Locate the "Connectors" section. 3. Find the custom connector your Owner added in the list (it will have a "Custom" label). 4. Click "Connect" to authenticate and start using the connector with Claude. ### For Pro and Max plans If you are using an individual Pro or Max plan, follow these steps to add a custom connector: 1. Navigate to **[Settings \> Connectors](https://claude.ai/settings/connectors)**. 2. Locate the "Connectors" section. 3. Click "Add custom connector" at the bottom of the section. 4. Add your connector's remote MCP server URL. 5. Optionally, click “Advanced settings” to specify an OAuth Client ID and OAuth Client Secret for your server. 6. Finish configuring your connector by clicking "Add." ### Enabling connectors after configuration You can enable connectors for individual conversations via the “+” button on the lower left of your chat interface, then "Connectors." You'll see your configured connectors with toggles allowing you to enable/disable them per conversation. ## How to remove connectors You can remove a custom connector by following these steps: 1. Navigate to **[Settings \> Connectors](https://claude.ai/settings/connectors)** 1. Team and Enterprise Owners can do this on their organization's behalf in **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)** 2. Locate the "Connectors" section. 3. Click "Remove" or select the three dots next to the connector you'd like to remove. 4. Follow the prompts to remove. If you're hoping to edit a custom connector, you'll need to remove it first, then re-add it using the updated details. ## How to build custom connectors To learn about building connectors to use with Claude, see Building custom connectors via remote MCP servers. ## Security and privacy considerations Custom connectors allow you to connect Claude to arbitrary services that have not been verified by Anthropic. When you connect Claude to external services, you're granting it the ability to access and potentially modify data within those services based on your permissions. It’s important to make sure you’re only connecting to remote MCP servers that you trust and that you’re aware of Claude’s interactions with web connectors. ### Security and permissions When you add a custom connector to Claude, you'll typically go through an OAuth authentication process to securely sign in to the application and grant specific permissions. This allows Claude to interact with the application on your behalf, without Claude ever seeing your actual password. You can revoke these permissions at any time by disconnecting the connector in Claude's settings or the third-party service's security settings. Remote MCP servers act as intermediaries between Claude and external applications. You should: - **Only connect to trusted servers:** Only connect Claude to servers built and hosted by organizations and applications you trust. - **Review requested permissions carefully:** During auth, review what permissions the MCP server is requesting to the application. Limit these scopes when possible and deny access if requested permissions seem unnecessary. - **Be aware of prompt injections:** Malicious MCP servers may include hidden instructions that try to make Claude perform unintended actions. Claude has built-in protections that attempt to block these attacks, but it's important to pay attention to tool inputs & outputs and connect only to trusted servers. - **Monitor changes in tool behavior:** Server developers may update tool behavior unexpectedly, leading to unintended or malicious behavior. ### Reporting malicious MCP servers If you become aware of a malicious MCP server, please report it to our **[vulnerability disclosure program](https://hackerone.com/anthropic-vdp/)**, and choose [`https://github.com/modelcontextprotocol`](https://github.com/modelcontextprotocol) as the Asset. ### Taking actions with tools Remote MCP servers give Claude tools it can invoke during your conversation. The developer of an MCP server can define what these tools do, including: - Reading data from connected applications. - Creating, modifying, or deleting data in connected applications. - Taking actions on behalf of the user. Claude can only access resources that you've given the server permission to access, but you should: - Be aware of any actions Claude is taking and that they have no destructive or unintended effects. - Review Claude's tool approval requests carefully and only click "Allow always" when using a server and tool that you trust to run unsupervised. - Using the "Search and tools" menu, disable any tools that aren't relevant to the current conversation or that you don't want Claude to be able to invoke. ### Interactive connectors Some connectors can display interactive interfaces directly within your Claude conversations. Instead of only returning text-based responses, these connectors can open live, interactive apps — like dashboards, task boards, or design tools — right in the chat. Interactive connectors appear in two ways: - **Inline cards:** Compact components embedded in the conversation, showing summaries, confirmations, or quick actions. - **Fullscreen view:** Immersive interfaces for complex interactions like data visualizations or document editing. The conversation composer remains available so you can continue chatting with Claude. You can interact with these connectors directly — filtering data, checking off tasks, adjusting settings — without leaving the conversation. Any actions you take within the interface use the same permissions you granted when connecting the tool. **Admin controls:** Team and Enterprise plan owners can disable specific tool calls that render interactive connectors within **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. ### Using Claude with Research **Note:** **[Advanced Research](https://claude.com/blog/integrations)** is not currently able to invoke tools from local MCP servers. Research allows Claude to deeply investigate queries by searching through hundreds of internal and external sources. During the research process, Claude can invoke tools from your connectors automatically without further approval. When using Research with custom connectors: - Disable any tools that can take write actions in external applications. - Review Claude’s approval request carefully and be aware of which tools you’re granting Claude permission to invoke. - Be mindful of the impact of Claude sending a large number of requests to your connectors. See **[Using Research on Claude](../15-Claude-AI-Features/using-research-on-claude.md)** for more information about this feature. --- ## getting-started-with-claude-in-slack-6e40ff610f.md # Get started with Claude in Slack Claude in Slack will be switched over to the new Claude Tag experience on August 3, 2026. To integrate Claude and Slack, use Claude Tag instead. Learn more **[about Claude Tag](https://claude.com/product/tag)**. You can now integrate Claude and Slack, giving you two ways to use them together: add Claude directly to your Slack workspace, or enable the Slack connector for your Claude apps. ## What is Claude in Slack? The Claude app is available to users on paid Slack plans. Slack admins must approve the Claude app before individual users can access it. It’s how we’ve brought Claude’s capabilities directly to Slack, bringing AI assistance into your team’s workspace. This integration allows you to work with Claude without leaving Slack through three convenient surfaces: **Direct message with Claude**: Start a private conversation with @Claude. **AI assistant panel**: Click the Claude icon in Slack's AI assistant header to open a panel on the right side of your Slack window, allowing you to access Claude from anywhere in the Slack app. **Thread participation**: Mention @Claude in any thread to get Claude's help with the conversation. All surfaces provide the same capabilities that you have enabled in Claude, including web search and connections to your integrated tools, allowing you to seamlessly integrate AI assistance into your existing workflow. **Note:** Team and Enterprise plan users with access to Claude Code on the web can also route coding tasks directly to Claude Code by mentioning @Claude. See [Using Claude in Slack](https://support.claude.com/en/articles/12461605-using-claude-in-slack#h_adda66b697) for details on this beta feature. ------------------------------------------------------------------------ ## Enable and install Claude in Slack ### For Slack admins 1. Go to the [Claude app in the Slack App Marketplace](https://slack.com/marketplace/A08SF47R6P4). 2. Click "Add to Slack" on the Claude app page. 3. Review and approve the app for your organization. 4. Choose whether to deploy org-wide or to specific workspaces. **To install across all workspaces:** 1. Navigate to your Slack management workspace at: [https://app.slack.com/manage/\/workspaces/all](https://app.slack.com/manage/%3CINSERT_SLACK_ID%3E/workspaces/all) - Find your enterprise's Slack ID using the appropriate lookup method 2. Navigate to **Integrations → Installed apps → Add to more workspaces**. 3. Toggle through all relevant workspaces where you'd like to enable Claude. ### For individual users Once your Slack admin has approved Claude (or if you're on a personal Slack plan): 1. Find Claude in your apps list (search for "Claude" if it's not immediately visible), or go to the Slack App Marketplace. 2. Click "Connect Account” to be prompted to connect your Claude account: 3. In the window that opens, select which organization you would like to connect with Claude for Slack. 4. Click “Authorize” to allow Claude in Slack to access your Claude chat account: 5. You should see a confirmation message upon successful connection: 6. After successful authentication, return to Slack. 7. Click “+ New Chat” to start a conversation with Claude, or @mention Claude in any Slack conversation to access its capabilities. **Tip**: Add Claude to your Slack header for quick access by clicking the three dots "..." at the top right and selecting "Add this app to header." **Enable Claude Code in Slack (beta)** Team and Enterprise plan users can route coding tasks from Slack to Claude Code on the web. To enable this capability: 1. A Claude Owner or Primary Owner must enable Claude Code on the web by navigating to **[Organization settings \> Claude Code](http://claude.ai/admin-settings/claude-code)**. 2. Individual users must have access to Claude Code on the web. Once enabled, mentioning @Claude for coding tasks will automatically create a Claude Code session. Learn more about **[using Claude Code in Slack](https://support.claude.com/en/articles/12461605-using-claude-in-slack#h_adda66b697)**. ------------------------------------------------------------------------ ## What is the Slack connector? The Slack connector is available for all paid plans (Pro, Max, Team, and Enterprise). Enabling the Slack connector allows Claude to search within your Slack workspace’s channels, direct messages, and shared files to pull relevant context into your conversations. Note that members of Team and Enterprise plan organizations will not see the option to enable the Slack connector individually until it’s enabled by an Owner. **Important:** You must install Claude in Slack before enabling and using the Slack connector. ## Enable the Slack connector ### Team and Enterprise owners 1. Log in to your Owner or Primary Owner account and click your initials in the lower left corner. 2. Navigate to **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. 3. Under “Connectors,” click the "Enable" button next to the Slack connector. 4. Users can then authenticate in their individual connector settings to begin using Slack in Claude. ### Individual Pro, Max, Team, and Enterprise users 1. Log in to your Claude account and click your initials in the lower left corner. 2. Navigate to **[Settings \> Connectors](http://claude.ai/settings/connectors)**. 3. Find the Slack connector and click “Connect.” 4. Click "Connect" to authenticate with the connector and start using Slack in Claude. ------------------------------------------------------------------------ ## Manage your Claude in Slack connections ### View Claude app connection status 1. Click on the Claude app in your Slack sidebar. 2. Go to the "Home" tab. 3. You'll see your connection status, including your connected account email and organization name. ### Disconnect the Claude app To disconnect your Claude account from Slack: 1. Go to the Claude Home tab in Slack. 2. Under **Disconnect Claude Account**, click the red "Disconnect" button. 3. Confirm the disconnection. Disconnecting will: - Remove the connection between your Claude account and Slack workspace. - Delete all past Claude conversations in Slack from Claude (within 30 days). - Preserve conversations in Slack, but Claude won't have awareness of them if you reconnect ### Disconnect the Slack connector You can also disconnect the Slack connector from your Claude settings (or you can enable/disable the connector for an individual chat): 1. Go to [](https://claude.ai/settings/integrations) claude.ai/settings/connectors 2. Find **Slack** in your list of Connectors. 3. Click the menu (...) and select "Disconnect." ------------------------------------------------------------------------ ## Privacy and data ### Data storage Your Slack conversations with Claude remain separate from your Claude history, keeping work organized across platforms. ### Data visibility - Conversations initiated in Slack are not visible in [your Claude chat history](http://claude.ai/recents). - Conversations initiated in the Claude web app are not accessible in Slack. - Each platform maintains separate conversation histories. ### Data deletion - Conversations are automatically deleted from Claude within 30 days if you disconnect the integration or uninstall the app. - Your conversations in Slack follow your organization's Slack retention policies. ------------------------------------------------------------------------ ## Frequently asked questions ### I’m trying to add Claude in Slack but it’s not working – help! If you are using a company Slack instance and are not assigned to an Admin role, a Slack Admin must approve the Claude app on behalf of your organization before you’re able to download it. If you try to skip this step and install Claude in Slack, you’ll see a **Request to install** prompt where you can send a message to your Slack Admin. Work with them to approve the app and make it available for your team. --- ## getting-started-with-claude-in-slack.md # Getting started with Claude in Slack You can now integrate Claude and Slack, giving you two ways to use them together: add Claude directly to your Slack workspace, or enable the Slack connector for your Claude apps. ## What is Claude in Slack? The Claude app is available to users on paid Slack plans. Slack admins must approve the Claude app before individual users can access it. It’s how we’ve brought Claude’s capabilities directly to Slack, bringing AI assistance into your team’s workspace. This integration allows you to work with Claude without leaving Slack through three convenient surfaces: **Direct message with Claude**: Start a private conversation with @Claude. **AI assistant panel**: Click the Claude icon in Slack's AI assistant header to open a panel on the right side of your Slack window, allowing you to access Claude from anywhere in the Slack app. **Thread participation**: Mention @Claude in any thread to get Claude's help with the conversation. All surfaces provide the same capabilities that you have enabled in Claude, including web search and connections to your integrated tools, allowing you to seamlessly integrate AI assistance into your existing workflow. **Note:** Team and Enterprise plan users with access to Claude Code on the web can also route coding tasks directly to Claude Code by mentioning @Claude. See [Using Claude in Slack](using-claude-in-slack.md#h_adda66b697) for details on this beta feature. ------------------------------------------------------------------------ ## Enabling and installing Claude in Slack ### For Slack admins 1. Go to the [Claude app in the Slack App Marketplace](https://slack.com/marketplace/A08SF47R6P4). 2. Click "Add to Slack" on the Claude app page. 3. Review and approve the app for your organization. 4. Choose whether to deploy org-wide or to specific workspaces. **To install across all workspaces:** 1. Navigate to your Slack management workspace at: [https://app.slack.com/manage/\/workspaces/all](https://app.slack.com/manage/%3CINSERT_SLACK_ID%3E/workspaces/all) - Find your enterprise's Slack ID using the appropriate lookup method 2. Navigate to **Integrations → Installed apps → Add to more workspaces**. 3. Toggle through all relevant workspaces where you'd like to enable Claude. ### For individual users Once your Slack admin has approved Claude (or if you're on a personal Slack plan): 1. Find Claude in your apps list (search for "Claude" if it's not immediately visible), or go to the Slack App Marketplace. 2. Click "Connect Account” to be prompted to connect your Claude account: 3. In the window that opens, select which organization you would like to connect with Claude for Slack. 4. Click “Authorize” to allow Claude in Slack to access your Claude chat account: 5. You should see a confirmation message upon successful connection: 6. After successful authentication, return to Slack. 7. Click “+ New Chat” to start a conversation with Claude, or @mention Claude in any Slack conversation to access its capabilities. **Tip**: Add Claude to your Slack header for quick access by clicking the three dots "..." at the top right and selecting "Add this app to header." **Enabling Claude Code in Slack (beta)** Team and Enterprise plan users can route coding tasks from Slack to Claude Code on the web. To enable this capability: 1. A Claude Owner or Primary Owner must enable Claude Code on the web by navigating to **[Organization settings \> Claude Code](http://claude.ai/admin-settings/claude-code)**. 2. Individual users must have access to Claude Code on the web. Once enabled, mentioning @Claude for coding tasks will automatically create a Claude Code session. Learn more about **[using Claude Code in Slack](using-claude-in-slack.md#h_adda66b697)**. ------------------------------------------------------------------------ ## What is the Slack connector? The Slack connector is available for all paid plans (Pro, Max, Team, and Enterprise). Enabling the Slack connector allows Claude to search within your Slack workspace’s channels, direct messages, and shared files to pull relevant context into your conversations. Note that members of Team and Enterprise plan organizations will not see the option to enable the Slack connector individually until it’s enabled by an Owner. **Important:** You must install Claude in Slack before enabling and using the Slack connector. ## Enabling the Slack connector ### Team and Enterprise owners 1. Log in to your Owner or Primary Owner account and click your initials in the lower left corner. 2. Navigate to **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. 3. Under “Connectors,” click the "Enable" button next to the Slack connector. 4. Users can then authenticate in their individual connector settings to begin using Slack in Claude. ### Individual Pro, Max, Team, and Enterprise users 1. Log in to your Claude account and click your initials in the lower left corner. 2. Navigate to **[Settings \> Connectors](http://claude.ai/settings/connectors)**. 3. Find the Slack connector and click “Connect.” 4. Click "Connect" to authenticate with the connector and start using Slack in Claude. ------------------------------------------------------------------------ ## Managing your Claude in Slack connections ### Viewing Claude app connection status 1. Click on the Claude app in your Slack sidebar. 2. Go to the "Home" tab. 3. You'll see your connection status, including your connected account email and organization name. ### Disconnecting the Claude app To disconnect your Claude account from Slack: 1. Go to the Claude Home tab in Slack. 2. Under **Disconnect Claude Account**, click the red "Disconnect" button. 3. Confirm the disconnection. Disconnecting will: - Remove the connection between your Claude account and Slack workspace. - Delete all past Claude conversations in Slack from Claude (within 30 days). - Preserve conversations in Slack, but Claude won't have awareness of them if you reconnect ### Disconnecting the Slack connector You can also disconnect the Slack connector from your Claude settings (or you can enable/disable the connector for an individual chat): 1. Go to [](https://claude.ai/settings/integrations) claude.ai/settings/connectors 2. Find **Slack** in your list of Connectors. 3. Click the menu (...) and select "Disconnect." ------------------------------------------------------------------------ ## Privacy and data ### Data storage Your Slack conversations with Claude remain separate from your Claude history, keeping work organized across platforms. ### Data visibility - Conversations initiated in Slack are not visible in [your Claude chat history](http://claude.ai/recents). - Conversations initiated in the Claude web app are not accessible in Slack. - Each platform maintains separate conversation histories. ### Data deletion - Conversations are automatically deleted from Claude within 30 days if you disconnect the integration or uninstall the app. - Your conversations in Slack follow your organization's Slack retention policies. ------------------------------------------------------------------------ ## Frequently asked questions ### I’m trying to add Claude in Slack but it’s not working – help! If you are using a company Slack instance and are not assigned to an Admin role, a Slack Admin must approve the Claude app on behalf of your organization before you’re able to download it. If you try to skip this step and install Claude in Slack, you’ll see a **Request to install** prompt where you can send a message to your Slack Admin. Work with them to approve the app and make it available for your team. --- ## getting-started-with-custom-connectors-using-remote-mcp-74392cf540.md # Get started with custom connectors using remote MCP Custom connectors using remote MCP are available on Claude, Cowork, and Claude Desktop for users on Free, Pro, Max, Team, and Enterprise plans. Free users are limited to one custom connector. ## What are custom connectors? Custom connectors let you connect Claude directly to the tools and data sources that matter most to your workflows. This enables Claude to operate within your favorite software and draw insights from the complete context of your external tools. You can: - Connect Claude to existing remote MCP servers. - Build your own remote MCP servers to connect with any tool. **Security and privacy with custom connectors** Custom connectors allow you to connect Claude to services that haven't been verified by Anthropic. Once connected, Claude can access those services and take action in them. For more guidance, review the **[Security and privacy considerations](#h_b79c05dfcd)** section below. ## What are remote MCP servers? The Model Context Protocol (MCP) is an open standard, created by Anthropic, for AI applications to connect to tools and data. Previously, **[MCP servers only ran locally](https://support.claude.com/en/articles/10949351-getting-started-with-local-mcp-servers-on-claude-desktop)** (i.e. on a user's laptop). Now, developers can build and host remote MCP servers that communicate with AI apps over the internet. Remote MCP servers give models access to internet-hosted tools and data, transforming Claude into an informed teammate that can independently handle complex, multi-step projects tailored to your needs. ------------------------------------------------------------------------ ## Network requirements When you add a custom connector, Claude connects to your remote MCP server from Anthropic's cloud infrastructure, rather than from your local device. This is true across every Claude client, including claude.ai, Claude Desktop, Cowork, and the mobile apps. This means your MCP server must be reachable over the public internet from Anthropic's IP ranges. Servers hosted on a private corporate network, behind a VPN, or blocked by a firewall won't connect, even if you can reach them from your own machine. ### If your server is on a private network You’ll need to allowlist Anthropic's IP addresses in your firewall so inbound connections from Claude can reach your server. See **[Anthropic IP addresses](https://platform.claude.com/docs/en/api/ip-addresses)** for the current ranges. ### Why this applies to Cowork and Claude Desktop Even though Cowork and Claude Desktop run on your computer, remote connectors are configured and brokered through your Claude account. The connection to your MCP server originates from Anthropic's servers, not from your machine's network interface. Local MCP servers configured in Claude Desktop via `claude_desktop_config.json` are a separate mechanism and do use your local network, but those aren't available in Cowork or claude.ai. ------------------------------------------------------------------------ ## Add a custom connector **Note:** While anyone can build and host connectors using remote MCP, only Owners can add them to Team and Enterprise plans. Once a connector has been added to a Team or Enterprise organization, users individually connect to and enable that connector. This ensures that Claude can only access tools and data that the individual user has access to. ### For Team and Enterprise plans **Preliminary steps for owners:** Before members of Team and Enterprise plans can configure custom connectors, an Owner or Primary Owner needs to follow these initial steps to add a custom connector to your organization: 1. Navigate to **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. 2. Click the "Add" button. 3. Hover over “Custom,” then select “Web.” 4. Add your connector's remote MCP server URL. 5. Optionally, click “Advanced settings” to specify an OAuth Client ID and OAuth Client Secret for your server. 6. Finish configuring your connector by clicking "Add." **Steps for members after connector is configured:** 1. Navigate to **[Customize \> Connectors](https://claude.ai/customize/connectors)** 2. Find the custom connector your Owner added in the list (it will have a "Custom" label). 3. Click "Connect" to authenticate and start using the connector with Claude. ### For Pro and Max plans If you are using an individual Pro or Max plan, follow these steps to add a custom connector: 1. Navigate to **[Customize \> Connectors](https://claude.ai/customize/connectors)** 2. Click "+" then “Add custom connector.” 3. Add your connector's remote MCP server URL. 4. Optionally, click “Advanced settings” to specify an OAuth Client ID and OAuth Client Secret for your server. 5. Finish configuring your connector by clicking "Add." ### Enabling connectors after configuration You can enable connectors for individual conversations via the “+” button on the lower left of your chat interface, then "Connectors." You'll see your configured connectors with toggles allowing you to enable/disable them per conversation. ------------------------------------------------------------------------ ## Remove custom connectors You can remove a custom connector by following these steps: 1. Navigate to **[Customize \> Connectors](https://claude.ai/customize/connectors)** 1. Team and Enterprise Owners can do this on their organization's behalf in **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)** 2. Locate the "Connectors" section. 3. Click "Remove" or select the three dots next to the connector you'd like to remove. 4. Follow the prompts to remove. If you're hoping to edit a custom connector, you'll need to remove it first, then re-add it using the updated details. ------------------------------------------------------------------------ ## Build custom connectors To learn about building connectors to use with Claude, see **[Building custom connectors via remote MCP servers](https://support.claude.com/en/articles/11503834-building-custom-connectors-via-remote-mcp-servers)**. ------------------------------------------------------------------------ ## Security and privacy considerations Custom connectors allow you to connect Claude to arbitrary services that have not been verified by Anthropic. When you connect Claude to external services, you're granting it the ability to access and potentially modify data within those services based on your permissions. It’s important to make sure you’re only connecting to remote MCP servers that you trust and that you’re aware of Claude’s interactions with web connectors. ### Security and permissions When you add a custom connector to Claude, you'll typically go through an OAuth authentication process to securely sign in to the application and grant specific permissions. This allows Claude to interact with the application on your behalf, without Claude ever seeing your actual password. You can revoke these permissions at any time by disconnecting the connector in Claude's settings or the third-party service's security settings. Remote MCP servers act as intermediaries between Claude and external applications. You should: - **Only connect to trusted servers:** Only connect Claude to servers built and hosted by organizations and applications you trust. - **Review requested permissions carefully:** During auth, review what permissions the MCP server is requesting to the application. Limit these scopes when possible and deny access if requested permissions seem unnecessary. - **Be aware of prompt injections:** Malicious MCP servers may include hidden instructions that try to make Claude perform unintended actions. Claude has built-in protections that attempt to block these attacks, but it's important to pay attention to tool inputs & outputs and connect only to trusted servers. - **Monitor changes in tool behavior:** Server developers may update tool behavior unexpectedly, leading to unintended or malicious behavior. ### Reporting malicious MCP servers If you become aware of a malicious MCP server, please report it to our **[vulnerability disclosure program](https://hackerone.com/anthropic-vdp/)**, and choose [`https://github.com/modelcontextprotocol`](https://github.com/modelcontextprotocol) as the Asset. ### Taking actions with tools Remote MCP servers give Claude tools it can invoke during your conversation. The developer of an MCP server can define what these tools do, including: - Reading data from connected applications. - Creating, modifying, or deleting data in connected applications. - Taking actions on behalf of the user. Claude can only access resources that you've given the server permission to access, but you should: - Be aware of any actions Claude is taking and that they have no destructive or unintended effects. - Review Claude's tool approval requests carefully and only click "Allow always" when using a server and tool that you trust to run unsupervised. - Using the "Search and tools" menu, disable any tools that aren't relevant to the current conversation or that you don't want Claude to be able to invoke. ### Interactive connectors Some connectors can display interactive interfaces directly within your Claude conversations. Instead of only returning text-based responses, these connectors can open live, interactive apps — like dashboards, task boards, or design tools — right in the chat. Interactive connectors appear in two ways: - **Inline cards:** Compact components embedded in the conversation, showing summaries, confirmations, or quick actions. - **Fullscreen view:** Immersive interfaces for complex interactions like data visualizations or document editing. The conversation composer remains available so you can continue chatting with Claude. You can interact with these connectors directly — filtering data, checking off tasks, adjusting settings — without leaving the conversation. Any actions you take within the interface use the same permissions you granted when connecting the tool. **Admin controls:** Team and Enterprise plan owners can disable specific tool calls that render interactive connectors within **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. ### Using Claude with Research **Note: [Advanced Research](https://claude.com/blog/integrations)** is not currently able to invoke tools from local MCP servers. Research allows Claude to deeply investigate queries by searching through hundreds of internal and external sources. During the research process, Claude can invoke tools from your connectors automatically without further approval. When using Research with custom connectors: - Disable any tools that can take write actions in external applications. - Review Claude’s approval request carefully and be aware of which tools you’re granting Claude permission to invoke. - Be mindful of the impact of Claude sending a large number of requests to your connectors. See **[Using Research on Claude](https://support.claude.com/en/articles/11088861-using-research-on-claude)** for more information about this feature. --- ## getting-started-with-custom-connectors-using-remote-mcp.md # Get started with custom connectors using remote MCP Custom connectors using remote MCP are available on Claude and Claude Desktop for users on free, Pro, Max, Team, and Enterprise plans. Free users are limited to one custom connector. This feature is currently in beta. ## What are custom connectors? Custom connectors let you connect Claude directly to the tools and data sources that matter most to your workflows. This enables Claude to operate within your favorite software and draw insights from the complete context of your external tools. You can: - Connect Claude to existing remote MCP servers. - Build your own remote MCP servers to connect with any tool. **⚠️ Security and Privacy with Custom Connectors (beta)** Be aware that custom connectors allow you to connect Claude to services that have not been verified by Anthropic, and allow Claude to access and take action in these services. For more guidance, review the **[Security and privacy considerations](#h_9088ccdf4d)** section below. ## What are remote MCP servers? The Model Context Protocol (MCP) is an open standard, created by Anthropic, for AI applications to connect to tools and data. Previously, **[MCP servers only ran locally](https://support.claude.com/en/articles/10949351-getting-started-with-local-mcp-servers-on-claude-desktop)** (i.e. on a user's laptop). Now, developers can build and host remote MCP servers that communicate with AI apps over the internet. Remote MCP servers give models access to internet-hosted tools and data, transforming Claude into an informed teammate that can independently handle complex, multi-step projects tailored to your needs. ## Add a custom connector **Note:** While anyone can build and host connectors using remote MCP, only Owners can add them to Team and Enterprise plans. Once a connector has been added to a Team or Enterprise organization, users individually connect to and enable that connector. This ensures that Claude can only access tools and data that the individual user has access to. ### For Team and Enterprise plans **Preliminary steps for owners:** Before members of Team and Enterprise plans can configure custom connectors, an Owner or Primary Owner needs to follow these initial steps to add a custom connector to your organization: 1. Navigate to **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. 2. Click "Add custom connector" at the bottom of the section. 3. Add your connector's remote MCP server URL. 4. Optionally, click “Advanced settings” to specify an OAuth Client ID and OAuth Client Secret for your server. 5. Finish configuring your connector by clicking "Add." **Steps for members after connector is configured:** 1. Navigate to **[Settings \> Connectors](https://claude.ai/settings/connectors)**. 2. Locate the "Connectors" section. 3. Find the custom connector your Owner added in the list (it will have a "Custom" label). 4. Click "Connect" to authenticate and start using the connector with Claude. ### For Pro and Max plans If you are using an individual Pro or Max plan, follow these steps to add a custom connector: 1. Navigate to **[Settings \> Connectors](https://claude.ai/settings/connectors)**. 2. Locate the "Connectors" section. 3. Click "Add custom connector" at the bottom of the section. 4. Add your connector's remote MCP server URL. 5. Optionally, click “Advanced settings” to specify an OAuth Client ID and OAuth Client Secret for your server. 6. Finish configuring your connector by clicking "Add." ### Enabling connectors after configuration You can enable connectors for individual conversations via the “+” button on the lower left of your chat interface, then "Connectors." You'll see your configured connectors with toggles allowing you to enable/disable them per conversation. ## How to remove connectors You can remove a custom connector by following these steps: 1. Navigate to **[Settings \> Connectors](https://claude.ai/settings/connectors)** 1. Team and Enterprise Owners can do this on their organization's behalf in **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)** 2. Locate the "Connectors" section. 3. Click "Remove" or select the three dots next to the connector you'd like to remove. 4. Follow the prompts to remove. If you're hoping to edit a custom connector, you'll need to remove it first, then re-add it using the updated details. ## How to build custom connectors To learn about building connectors to use with Claude, see Building custom connectors via remote MCP servers. ## Security and privacy considerations Custom connectors allow you to connect Claude to arbitrary services that have not been verified by Anthropic. When you connect Claude to external services, you're granting it the ability to access and potentially modify data within those services based on your permissions. It’s important to make sure you’re only connecting to remote MCP servers that you trust and that you’re aware of Claude’s interactions with web connectors. ### Security and permissions When you add a custom connector to Claude, you'll typically go through an OAuth authentication process to securely sign in to the application and grant specific permissions. This allows Claude to interact with the application on your behalf, without Claude ever seeing your actual password. You can revoke these permissions at any time by disconnecting the connector in Claude's settings or the third-party service's security settings. Remote MCP servers act as intermediaries between Claude and external applications. You should: - **Only connect to trusted servers:** Only connect Claude to servers built and hosted by organizations and applications you trust. - **Review requested permissions carefully:** During auth, review what permissions the MCP server is requesting to the application. Limit these scopes when possible and deny access if requested permissions seem unnecessary. - **Be aware of prompt injections:** Malicious MCP servers may include hidden instructions that try to make Claude perform unintended actions. Claude has built-in protections that attempt to block these attacks, but it's important to pay attention to tool inputs & outputs and connect only to trusted servers. - **Monitor changes in tool behavior:** Server developers may update tool behavior unexpectedly, leading to unintended or malicious behavior. ### Reporting malicious MCP servers If you become aware of a malicious MCP server, please report it to our **[vulnerability disclosure program](https://hackerone.com/anthropic-vdp/)**, and choose [`https://github.com/modelcontextprotocol`](https://github.com/modelcontextprotocol) as the Asset. ### Taking actions with tools Remote MCP servers give Claude tools it can invoke during your conversation. The developer of an MCP server can define what these tools do, including: - Reading data from connected applications. - Creating, modifying, or deleting data in connected applications. - Taking actions on behalf of the user. Claude can only access resources that you've given the server permission to access, but you should: - Be aware of any actions Claude is taking and that they have no destructive or unintended effects. - Review Claude's tool approval requests carefully and only click "Allow always" when using a server and tool that you trust to run unsupervised. - Using the "Search and tools" menu, disable any tools that aren't relevant to the current conversation or that you don't want Claude to be able to invoke. ### Interactive connectors Some connectors can display interactive interfaces directly within your Claude conversations. Instead of only returning text-based responses, these connectors can open live, interactive apps — like dashboards, task boards, or design tools — right in the chat. Interactive connectors appear in two ways: - **Inline cards:** Compact components embedded in the conversation, showing summaries, confirmations, or quick actions. - **Fullscreen view:** Immersive interfaces for complex interactions like data visualizations or document editing. The conversation composer remains available so you can continue chatting with Claude. You can interact with these connectors directly — filtering data, checking off tasks, adjusting settings — without leaving the conversation. Any actions you take within the interface use the same permissions you granted when connecting the tool. **Admin controls:** Team and Enterprise plan owners can disable specific tool calls that render interactive connectors within **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. ### Using Claude with Research **Note:** **[Advanced Research](https://claude.com/blog/integrations)** is not currently able to invoke tools from local MCP servers. Research allows Claude to deeply investigate queries by searching through hundreds of internal and external sources. During the research process, Claude can invoke tools from your connectors automatically without further approval. When using Research with custom connectors: - Disable any tools that can take write actions in external applications. - Review Claude’s approval request carefully and be aware of which tools you’re granting Claude permission to invoke. - Be mindful of the impact of Claude sending a large number of requests to your connectors. See **[Using Research on Claude](../15-Claude-AI-Features/using-research-on-claude.md)** for more information about this feature. --- ## getting-started-with-custom-integrations-using-remote-mcp.md # Get started with custom connectors using remote MCP Custom connectors using remote MCP are available on Claude, Cowork, and Claude Desktop for users on free, Pro, Max, Team, and Enterprise plans. Free users are limited to one custom connector. This feature is currently in beta. ## What are custom connectors? Custom connectors let you connect Claude directly to the tools and data sources that matter most to your workflows. This enables Claude to operate within your favorite software and draw insights from the complete context of your external tools. You can: - Connect Claude to existing remote MCP servers. - Build your own remote MCP servers to connect with any tool. **⚠️ Security and Privacy with Custom Connectors (beta)** Be aware that custom connectors allow you to connect Claude to services that have not been verified by Anthropic, and allow Claude to access and take action in these services. For more guidance, review the **[Security and privacy considerations](#h_9088ccdf4d)** section below. ## What are remote MCP servers? The Model Context Protocol (MCP) is an open standard, created by Anthropic, for AI applications to connect to tools and data. Previously, **[MCP servers only ran locally](https://support.claude.com/en/articles/10949351-getting-started-with-local-mcp-servers-on-claude-desktop)** (i.e. on a user's laptop). Now, developers can build and host remote MCP servers that communicate with AI apps over the internet. Remote MCP servers give models access to internet-hosted tools and data, transforming Claude into an informed teammate that can independently handle complex, multi-step projects tailored to your needs. ## Add a custom connector **Note:** While anyone can build and host connectors using remote MCP, only Owners can add them to Team and Enterprise plans. Once a connector has been added to a Team or Enterprise organization, users individually connect to and enable that connector. This ensures that Claude can only access tools and data that the individual user has access to. ### For Team and Enterprise plans **Preliminary steps for owners:** Before members of Team and Enterprise plans can configure custom connectors, an Owner or Primary Owner needs to follow these initial steps to add a custom connector to your organization: 1. Navigate to **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. 2. Click "Add custom connector" at the bottom of the section. 3. Add your connector's remote MCP server URL. 4. Optionally, click “Advanced settings” to specify an OAuth Client ID and OAuth Client Secret for your server. 5. Finish configuring your connector by clicking "Add." **Steps for members after connector is configured:** 1. Navigate to **[Settings \> Connectors](https://claude.ai/settings/connectors)**. 2. Locate the "Connectors" section. 3. Find the custom connector your Owner added in the list (it will have a "Custom" label). 4. Click "Connect" to authenticate and start using the connector with Claude. ### For Pro and Max plans If you are using an individual Pro or Max plan, follow these steps to add a custom connector: 1. Navigate to **[Settings \> Connectors](https://claude.ai/settings/connectors)**. 2. Locate the "Connectors" section. 3. Click "Add custom connector" at the bottom of the section. 4. Add your connector's remote MCP server URL. 5. Optionally, click “Advanced settings” to specify an OAuth Client ID and OAuth Client Secret for your server. 6. Finish configuring your connector by clicking "Add." ### Enabling connectors after configuration You can enable connectors for individual conversations via the “+” button on the lower left of your chat interface, then "Connectors." You'll see your configured connectors with toggles allowing you to enable/disable them per conversation. ## How to remove connectors You can remove a custom connector by following these steps: 1. Navigate to **[Settings \> Connectors](https://claude.ai/settings/connectors)** 1. Team and Enterprise Owners can do this on their organization's behalf in **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)** 2. Locate the "Connectors" section. 3. Click "Remove" or select the three dots next to the connector you'd like to remove. 4. Follow the prompts to remove. If you're hoping to edit a custom connector, you'll need to remove it first, then re-add it using the updated details. ## How to build custom connectors To learn about building connectors to use with Claude, see Building custom connectors via remote MCP servers. ## Security and privacy considerations Custom connectors allow you to connect Claude to arbitrary services that have not been verified by Anthropic. When you connect Claude to external services, you're granting it the ability to access and potentially modify data within those services based on your permissions. It’s important to make sure you’re only connecting to remote MCP servers that you trust and that you’re aware of Claude’s interactions with web connectors. ### Security and permissions When you add a custom connector to Claude, you'll typically go through an OAuth authentication process to securely sign in to the application and grant specific permissions. This allows Claude to interact with the application on your behalf, without Claude ever seeing your actual password. You can revoke these permissions at any time by disconnecting the connector in Claude's settings or the third-party service's security settings. Remote MCP servers act as intermediaries between Claude and external applications. You should: - **Only connect to trusted servers:** Only connect Claude to servers built and hosted by organizations and applications you trust. - **Review requested permissions carefully:** During auth, review what permissions the MCP server is requesting to the application. Limit these scopes when possible and deny access if requested permissions seem unnecessary. - **Be aware of prompt injections:** Malicious MCP servers may include hidden instructions that try to make Claude perform unintended actions. Claude has built-in protections that attempt to block these attacks, but it's important to pay attention to tool inputs & outputs and connect only to trusted servers. - **Monitor changes in tool behavior:** Server developers may update tool behavior unexpectedly, leading to unintended or malicious behavior. ### Reporting malicious MCP servers If you become aware of a malicious MCP server, please report it to our **[vulnerability disclosure program](https://hackerone.com/anthropic-vdp/)**, and choose [`https://github.com/modelcontextprotocol`](https://github.com/modelcontextprotocol) as the Asset. ### Taking actions with tools Remote MCP servers give Claude tools it can invoke during your conversation. The developer of an MCP server can define what these tools do, including: - Reading data from connected applications. - Creating, modifying, or deleting data in connected applications. - Taking actions on behalf of the user. Claude can only access resources that you've given the server permission to access, but you should: - Be aware of any actions Claude is taking and that they have no destructive or unintended effects. - Review Claude's tool approval requests carefully and only click "Allow always" when using a server and tool that you trust to run unsupervised. - Using the "Search and tools" menu, disable any tools that aren't relevant to the current conversation or that you don't want Claude to be able to invoke. ### Interactive connectors Some connectors can display interactive interfaces directly within your Claude conversations. Instead of only returning text-based responses, these connectors can open live, interactive apps — like dashboards, task boards, or design tools — right in the chat. Interactive connectors appear in two ways: - **Inline cards:** Compact components embedded in the conversation, showing summaries, confirmations, or quick actions. - **Fullscreen view:** Immersive interfaces for complex interactions like data visualizations or document editing. The conversation composer remains available so you can continue chatting with Claude. You can interact with these connectors directly — filtering data, checking off tasks, adjusting settings — without leaving the conversation. Any actions you take within the interface use the same permissions you granted when connecting the tool. **Admin controls:** Team and Enterprise plan owners can disable specific tool calls that render interactive connectors within **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. ### Using Claude with Research **Note:** **[Advanced Research](https://claude.com/blog/integrations)** is not currently able to invoke tools from local MCP servers. Research allows Claude to deeply investigate queries by searching through hundreds of internal and external sources. During the research process, Claude can invoke tools from your connectors automatically without further approval. When using Research with custom connectors: - Disable any tools that can take write actions in external applications. - Review Claude’s approval request carefully and be aware of which tools you’re granting Claude permission to invoke. - Be mindful of the impact of Claude sending a large number of requests to your connectors. See **[Using Research on Claude](../15-Claude-AI-Features/using-research-on-claude.md)** for more information about this feature. --- ## google-sheets-add-on.md # Google Sheets add-on The [Claude for Sheets extension](https://workspace.google.com/marketplace/app/claude%5Ffor%5Fsheets/909417792257) integrates Claude into Google Sheets, allowing you to execute interactions with Claude directly in cells. ## Why use Claude for Sheets? Claude for Sheets enables prompt engineering at scale by enabling you to test prompts across evaluation suites in parallel. Additionally, it excels at office tasks like survey analysis and online data processing. Visit our [prompt engineering example sheet](https://docs.google.com/spreadsheets/d/1sUrBWO0u1-ZuQ8m5gt3-1N5PLR6r__UsRsB7WeySDQA/copy) to see this in action. ## Get started with Claude for Sheets ### Install Claude for Sheets Easily enable Claude for Sheets using the following steps: 1. **Get your Claude API key:** If you don't yet have an API key, you can make API keys in the Claude Console. 2. **Install the Claude for Sheets extension:** Find the [Claude for Sheets extension](https://workspace.google.com/marketplace/app/claude%5Ffor%5Fsheets/909417792257) in the add-on marketplace, then click the blue \`Install\` button and accept the permissions. 1. Permissions: The Claude for Sheets extension will ask for a variety of permissions needed to function properly. Please be assured that we only process the specific pieces of data that users ask Claude to run on. This data is never used to train our generative models. 2. Extension permissions include: 1. **View and manage spreadsheets that this application has been installed in:** Needed to run prompts and return results. 2. **Connect to an external service:** Needed in order to make calls to Claude API endpoints. 3. **Allow this application to run when you are not present:** Needed to run cell recalculations without user intervention. 4. **Display and run third-party web content in prompts and sidebars inside Google applications:** Needed to display the sidebar and post-install prompt. 3. **Connect your API key:** Enter your API key at `Extensions` \> `Claude for Sheets™` \> `Open sidebar` \> `☰` \> `Settings` \> `API provider`. You may need to wait or refresh for the Claude for Sheets menu to appear. **Important:** You will have to re-enter your API key every time you make a new Google Sheet. ### Enter your first prompt There are two main functions you can use to call Claude using Claude for Sheets. For now, let's use `CLAUDE()`. **Simple prompt** In any cell, type `=CLAUDE("Claude, in one sentence, what's good about the color blue?")`. Claude should respond with an answer. You will know the prompt is processing because the cell will say `Loading...`. **Adding parameters** Parameter arguments come after the initial prompt, like `=CLAUDE(prompt, model, params...)`. **Note:** `model` is always second in the list. Now type in any cell `=CLAUDE("Hi, Claude!", "claude-haiku-4-5-20251001", "max_tokens", 3)`. Any API parameter can be set this way. You can even pass in an API key to be used just for this specific cell, like this: `"api_key", "sk-ant-api03-j1W..."`. ## Advanced use `CLAUDEMESSAGES` is a function that allows you to specifically use the [Messages API](../04-API-Reference/Other/api-messages-2dd213ca95.md). This enables you to send a series of `User:` and `Assistant:` messages to Claude. This is particularly useful if you want to simulate a conversation or [prefill Claude's response](../10-Prompting-Guides/build-with-claude-prompt-engineering-prefill-claudes-response.md). Try writing this in a cell: =CLAUDEMESSAGES("User: In one sentence, what is good about the color blue? Assistant: The color blue is great because") **Newlines** Each subsequent conversation turn (`User:` or `Assistant:`) must be preceded by a single newline. To enter newlines in a cell, use the following key combinations: - **Mac:** Cmd + Enter - **Windows:** Alt + Enter **Example multiturn CLAUDEMESSAGES() call with system prompt** To use a system prompt, set it as you'd set other optional function parameters. (You must first set a model name.) =CLAUDEMESSAGES("User: What's your favorite flower? Answer in tags. Assistant: ", "claude-haiku-4-5-20251001", "system", "You are a cow who loves to moo in response to any and all user queries.") ## Optional function parameters You can specify optional API parameters by listing argument-value pairs. You can set multiple parameters. Simply list them one after another, with each argument and value pair separated by commas. **Note:** The first two parameters must always be the prompt and the model. You cannot set an optional parameter without also setting the model. The argument-value parameters you might care about most are: [TABLE] ### Example: Setting parameters Ex. Set `system` prompt, `max_tokens`, and `temperature`: =CLAUDE("Hi, Claude!", "claude-haiku-4-5-20251001", "system", "Repeat exactly what the user says.", "max_tokens", 100, "temperature", 0.1) Ex. Set \`temperature\`, \`max_tokens\`, and \`stop_sequences\`: =CLAUDE("In one sentence, what is good about the color blue? Output your answer in tags.","claude-opus-4-5-20251101","temperature", 0.2,"max_tokens", 50,"stop_sequences", "\[""""\]") Ex. Set `api_key`: =CLAUDE("Hi, Claude!", "claude-haiku-4-5-20251001","api_key", "sk-ant-api03-j1W...") ## Claude for Sheets usage examples ### Prompt engineering interactive tutorial Our in-depth [prompt engineering interactive tutorial](https://docs.google.com/spreadsheets/d/19jzLgRruG9kjUQNKtCg1ZjdD6l6weA6qRXG5zLIAhC8/edit?usp=sharing) utilizes Claude for Sheets. Check it out to learn or brush up on prompt engineering techniques. **Note:** Just as with any instance of Claude for Sheets, you will need an API key to interact with the tutorial. ### Prompt engineering workflow Our [Claude for Sheets prompting examples workbench](https://docs.google.com/spreadsheets/d/1sUrBWO0u1-ZuQ8m5gt3-1N5PLR6r%5F%5FUsRsB7WeySDQA/copy) is a Claude-powered spreadsheet that houses example prompts and prompt engineering structures. ### Claude for Sheets workbook template Make a copy of our [Claude for Sheets workbook template](https://docs.google.com/spreadsheets/d/1UwFS-ZQWvRqa6GkbL4sy0ITHK2AhXKe-jpMLzS0kTgk/copy) to get started with your own Claude for Sheets work! ## Troubleshooting ### NAME? Error: Unknown function: 'claude' 1. Ensure that you have enabled the extension for use in the current sheet 1. Go to *Extensions \> Add-ons \> Manage add-ons* 2. Click on the triple dot menu at the top right corner of the Claude for Sheets extension and make sure "Use in this document" is checked 2. Refresh the page ### \#ERROR!, ⚠ DEFERRED ⚠ or ⚠ THROTTLED ⚠ You can manually recalculate `#ERROR!`, `⚠ DEFERRED ⚠`, or `⚠ THROTTLED ⚠` cells by selecting from the recalculate options within the Claude for Sheets extension menu. ### Can't enter API key 1. Wait 20 seconds, then check again 2. Refresh the page and wait 20 seconds again 3. Uninstall and reinstall the extension ## Further information For more information regarding this extension, see the [Claude for Sheets Google Workspace Marketplace](https://workspace.google.com/marketplace/app/claude%5Ffor%5Fsheets/909417792257) overview page. --- ## intro-to-connectors-c3f720c004.md # Getting started with connectors Learn how to supercharge Claude by connecting the tools you already use. This tutorial shows you how to set up connectors that give Claude access to your files, apps, and workflows. - Professional - Claude.ai - Watch time 5 min 4 min - [Copy link](#) https://claude.com/resources/tutorials/getting-started-with-connectors ## Related tutorials [Delegating and scheduling tasks in Claude Cowork](/resources/tutorials/delegating-and-scheduling-tasks-in-claude-cowork) Delegating and scheduling tasks in Claude Cowork Delegating and scheduling tasks in Claude Cowork Tutorial [Tutorial](/resources/tutorials/delegating-and-scheduling-tasks-in-claude-cowork) Tutorial [Tasks to try with @Claude in your workspace](/resources/tutorials/tasks-to-try-with-claude-tag-in-your-workspace) Tasks to try with @Claude in your workspace Tasks to try with @Claude in your workspace Tutorial [Tutorial](/resources/tutorials/tasks-to-try-with-claude-tag-in-your-workspace) Tutorial [Best practices for using @Claude](/resources/tutorials/best-practices-using-claude-tag) Best practices for using @Claude Best practices for using @Claude Tutorial [Tutorial](/resources/tutorials/best-practices-using-claude-tag) Tutorial [Using Claude Cowork for legal: answer fast questions on past decisions](/resources/tutorials/using-claude-cowork-for-legal-question-briefing) Using Claude Cowork for legal: answer fast questions on past decisions Using Claude Cowork for legal: answer fast questions on past decisions Tutorial [Tutorial](/resources/tutorials/using-claude-cowork-for-legal-question-briefing) Tutorial [Homepage](https://claude.com) Homepage Thank you! Your submission has been received! Oops! Something went wrong while submitting the form. Write [Button Text](#) Button Text Learn [Button Text](#) Button Text Code [Button Text](#) Button Text Write - Help me develop a unique voice for an audience Hi Claude! Could you help me develop a unique voice for an audience? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Improve my writing style Hi Claude! Could you improve my writing style? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Brainstorm creative ideas Hi Claude! Could you brainstorm creative ideas? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! Learn - Explain a complex topic simply Hi Claude! Could you explain a complex topic simply? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Help me make sense of these ideas Hi Claude! Could you help me make sense of these ideas? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Prepare for an exam or interview Hi Claude! Could you prepare for an exam or interview? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! Code - Explain a programming concept Hi Claude! Could you explain a programming concept? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Look over my code and give me tips Hi Claude! Could you look over my code and give me tips? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Vibe code with me Hi Claude! Could you vibe code with me? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! More - Write case studies This is another test - Write grant proposals Hi Claude! Could you write grant proposals? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to — like Google Drive, web search, etc. — if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational. Please execute the task as soon as you can - an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Write video scripts this is a test [Anthropic](https://www.anthropic.com/) Anthropic © \[year\] Anthropic PBC Products - Claude [Claude](/product/overview) Claude - Claude Code [Claude Code](/product/claude-code) Claude Code - Claude Code for Enterprise [Claude Code for Enterprise](/product/claude-code/enterprise) Claude Code for Enterprise - Claude Cowork [Claude Cowork](/product/cowork) Claude Cowork - @Claude [@Claude](/product/tag) @Claude - Claude Design [Claude Design](/product/design) Claude Design - Claude Science [Claude Science](/product/claude-science) Claude Science - Claude Security [Claude Security](/product/claude-security) Claude Security - Download app [Download app](/download) Download app - Pricing [Pricing](/pricing) Pricing - Log in [Log in](https://claude.ai/login) Features - Claude for Chrome [Claude for Chrome](/claude-for-chrome) Claude for Chrome - Claude for Microsoft 365 [Claude for Microsoft 365](/claude-for-microsoft-365) Claude for Microsoft 365 - Skills [Skills](/skills) Skills Models - Mythos [Mythos](https://www.anthropic.com/claude/mythos) Mythos - Fable [Fable](https://www.anthropic.com/claude/fable) Fable - Opus [Opus](https://www.anthropic.com/claude/opus) Opus - Sonnet [Sonnet](https://www.anthropic.com/claude/sonnet) Sonnet - Haiku [Haiku](https://www.anthropic.com/claude/haiku) Haiku Solutions - AI agents [AI agents](/solutions/agents) AI agents - Code modernization [Code modernization](/solutions/code-modernization) Code modernization - Coding [Coding](/solutions/coding) Coding - Customer support [Customer support](/solutions/customer-support) Customer support - Cybersecurity [Cybersecurity](/solutions/cybersecurity) Cybersecurity - Enterprise [Enterprise](/solutions/enterprise) Enterprise - Financial services [Financial services](/solutions/financial-services) Financial services - Government [Government](/solutions/government) Government - Healthcare [Healthcare](/solutions/healthcare) Healthcare - Higher education [Higher education](/solutions/education) Higher education - K-12 teachers [K-12 teachers](/solutions/teachers) K-12 teachers - Legal [Legal](/solutions/legal) Legal - Life sciences [Life sciences](/solutions/life-sciences) Life sciences - Nonprofits [Nonprofits](/solutions/nonprofits) Nonprofits - Small business [Small business](/solutions/small-business) Small business Claude Platform - Overview [Overview](/platform/api) Overview - Developer docs [Developer docs](https://platform.claude.com/docs) Developer docs - Pricing [Pricing](https://claude.com/pricing#api) Pricing - Ecosystem [Ecosystem](/ecosystem) Ecosystem - Marketplace [Marketplace](/platform/marketplace) Marketplace - Claude on AWS [Claude on AWS](/partners/claude-on-aws) Claude on AWS - Google Cloud [Google Cloud](/partners/google-cloud) Google Cloud - Microsoft Foundry [Microsoft Foundry](/partners/microsoft-foundry) Microsoft Foundry - Regional compliance [Regional compliance](/regional-compliance) Regional compliance - Console login [Console login](https://platform.claude.com/) Console login Resources - Blog [Blog](/blog) Blog - Claude partner network [Claude partner network](/partners) Claude partner network - Community [Community](/community) Community - Connectors [Connectors](/connectors) Connectors - Courses [Courses](https://www.anthropic.com/learn) Courses - Customer stories [Customer stories](/customers) Customer stories - Engineering at Anthropic [Engineering at Anthropic](https://www.anthropic.com/engineering) Engineering at Anthropic - Events [Events](https://www.anthropic.com/events) Events - Plugins [Plugins](/plugins) Plugins - Powered by Claude [Powered by Claude](/partners/powered-by-claude) Powered by Claude - Service partners [Service partners](/partners/services) Service partners - Tutorials [Tutorials](/resources/tutorials) Tutorials - Use cases [Use cases](/resources/use-cases) Use cases Company - Anthropic [Anthropic](https://www.anthropic.com/) Anthropic - Careers [Careers](https://www.anthropic.com/careers) Careers - Policy [Policy](https://www.anthropic.com/policy) Policy - Economic Futures [Economic Futures](https://www.anthropic.com/economic-futures) Economic Futures - Research [Research](https://www.anthropic.com/research) Research - News [News](https://www.anthropic.com/news) News - Policy on the AI Exponential --- ## intro-to-connectors.md - [](../15-Claude-AI-Features/claude-opus-4-6-anthropic.md) Opus - [](../15-Claude-AI-Features/claude-sonnet-4-6-anthropic.md) Sonnet - [](../15-Claude-AI-Features/claude-haiku-4-5-anthropic.md) Haiku - Platform - [](/platform/api) Overview - [](https://platform.claude.com/docs) Developer docs - [](http://claude.com/pricing#api) Pricing - [](../04-API-Reference/Other/platform-claude-com.md) Console login - Solutions Use cases - [](/solutions/agents) AI agents - [](/solutions/claude-code-security) Claude Code Security - [](/solutions/coding) Coding Industries - [](/solutions/customer-support) Customer support - [](/solutions/education) Education - [](/solutions/financial-services) Financial services - [](/solutions/government) Government - [](/solutions/healthcare) Healthcare - [](/solutions/life-sciences) Life sciences - [](/solutions/nonprofits) Nonprofits - Pricing - [](/pricing) Overview - [](/pricing#api) API Pricing - [](/pricing/max) Max plan - [](/pricing/team) Team plan - [](/pricing/enterprise) Enterprise plan - Resources Insights - [](/blog) Blog - [](/customers) Customer stories - [](https://www.anthropic.com/news) Anthropic news Learn - [](https://www.anthropic.com/learn) Anthropic Academy - [](/resources/courses) Courses - [](/resources/tutorials) Tutorials - [](/resources/use-cases) Use cases Tools - [](/connectors) Connectors - [](/plugins) Plugins Connect - [](https://www.anthropic.com/events) Events - [](/community) Community - [](https://claude.ai/login) Login   - Contact sales [Contact sales](/contact-sales) Contact sales - Try Claude [Try Claude](https://claude.ai/) Try Claude - Contact sales [Contact sales](/contact-sales) Contact sales - Try Claude [Try Claude](https://claude.ai/) Try Claude [](#) - Contact sales [Contact sales](/contact-sales) Contact sales - Try Claude [Try Claude](https://claude.ai/) Try Claude - Contact sales [Contact sales](/contact-sales) Contact sales - Try Claude [Try Claude](https://claude.ai/) Try Claude - Meet Claude Products - [](/product/overview) Claude - [](/product/claude-code) Claude Code - [](/product/cowork) Claude Cowork Features - [](/claude-for-chrome) Claude for Chrome - [](/claude-for-slack) Claude for Slack - [](/claude-for-excel) Claude for Excel - [](/claude-for-powerpoint) Claude for PowerPoint - [](/skills) Skills Models - [](../15-Claude-AI-Features/claude-opus-4-6-anthropic.md) Opus - [](../15-Claude-AI-Features/claude-sonnet-4-6-anthropic.md) Sonnet - [](../15-Claude-AI-Features/claude-haiku-4-5-anthropic.md) Haiku - Platform - [](/platform/api) Overview - [](https://platform.claude.com/docs) Developer docs - [](http://claude.com/pricing#api) Pricing - [](../04-API-Reference/Other/platform-claude-com.md) Console login - Solutions Use cases - [](/solutions/agents) AI agents - [](/solutions/claude-code-security) Claude Code Security - [](/solutions/coding) Coding Industries - [](/solutions/customer-support) Customer support - [](/solutions/education) Education - [](/solutions/financial-services) Financial services - [](/solutions/government) Government - [](/solutions/healthcare) Healthcare - [](/solutions/life-sciences) Life sciences - [](/solutions/nonprofits) Nonprofits - Pricing - [](/pricing) Overview - [](/pricing#api) API Pricing - [](/pricing/max) Max plan - [](/pricing/team) Team plan - [](/pricing/enterprise) Enterprise plan - Resources Insights - [](/blog) Blog - [](/customers) Customer stories - [](https://www.anthropic.com/news) Anthropic news Learn - [](https://www.anthropic.com/learn) Anthropic Academy - [](/resources/courses) Courses - [](/resources/tutorials) Tutorials - [](/resources/use-cases) Use cases Tools - [](/connectors) Connectors - [](/plugins) Plugins Connect - [](https://www.anthropic.com/events) Events - [](/community) Community - [](https://claude.ai/login) Login - Contact sales [Contact sales](/contact-sales) Contact sales - Try Claude [Try Claude](https://claude.ai/) Try Claude - Contact sales [Contact sales](/contact-sales) Contact sales - Try Claude [Try Claude](https://claude.ai/) Try Claude 1. Tutorials [](/resources/tutorials) Tutorials / 2. Getting started with connectors Explore here - [](#) Ask questions about this page - [](#) Copy as markdown # Getting started with connectors Learn how to supercharge Claude by connecting the tools you already use. This tutorial shows you how to set up connectors that give Claude access to your files, apps, and workflows. - Category Professional - Product Claude.ai - Reading time Watch time 5 min 4 min - Share [Copy link](#) https://claude.com/resources/tutorials/getting-started-with-connectors ## Related tutorials [Choosing the right Claude model: Haiku, Sonnet, and Opus](/resources/tutorials/choosing-the-right-claude-model) Choosing the right Claude model: Haiku, Sonnet, and Opus Choosing the right Claude model: Haiku, Sonnet, and Opus Tutorial [Tutorial](/resources/tutorials/choosing-the-right-claude-model) Tutorial [Claude Enterprise Administrator Guide](/resources/tutorials/claude-enterprise-administrator-guide) Claude Enterprise Administrator Guide Claude Enterprise Administrator Guide Tutorial [Tutorial](/resources/tutorials/claude-enterprise-administrator-guide) Tutorial [How to customize plugins in Cowork](/resources/tutorials/how-to-customize-plugins-in-cowork) How to customize plugins in Cowork How to customize plugins in Cowork Tutorial [Tutorial](/resources/tutorials/how-to-customize-plugins-in-cowork) Tutorial [How to build a plugin from scratch in Cowork](/resources/tutorials/how-to-build-a-plugin-from-scratch-in-cowork) How to build a plugin from scratch in Cowork How to build a plugin from scratch in Cowork Tutorial [Tutorial](/resources/tutorials/how-to-build-a-plugin-from-scratch-in-cowork) Tutorial [Homepage](https://claude.com) Homepage [Next](#) Next Thank you! Your submission has been received! Oops! Something went wrong while submitting the form. Write [Button Text](#) Button Text Learn [Button Text](#) Button Text Code [Button Text](#) Button Text Write - Help me develop a unique voice for an audience [](#) Hi Claude! Could you help me develop a unique voice for an audience? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational.\ \ Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Improve my writing style [](#) Hi Claude! Could you improve my writing style? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational.\ \ Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Brainstorm creative ideas [](#) Hi Claude! Could you brainstorm creative ideas? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational.\ \ Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! Learn - Explain a complex topic simply [](#) Hi Claude! Could you explain a complex topic simply? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational.\ \ Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Help me make sense of these ideas [](#) Hi Claude! Could you help me make sense of these ideas? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational.\ \ Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Prepare for an exam or interview [](#) Hi Claude! Could you prepare for an exam or interview? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational.\ \ Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! Code - Explain a programming concept [](#) Hi Claude! Could you explain a programming concept? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational.\ \ Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Look over my code and give me tips [](#) Hi Claude! Could you look over my code and give me tips? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational.\ \ Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Vibe code with me [](#) Hi Claude! Could you vibe code with me? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to— like Google Drive, web search, etc.—if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational.\ \ Please execute the task as soon as you can—an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! More - Write case studies [](#) This is another test - Write grant proposals [](#) Hi Claude! Could you write grant proposals? If you need more information from me, ask me 1-2 key questions right away. If you think I should upload any documents that would help you do a better job, let me know. You can use the tools you have access to — like Google Drive, web search, etc. — if they’ll help you better accomplish this task. Do not use analysis tool. Please keep your responses friendly, brief and conversational.\ \ Please execute the task as soon as you can - an artifact would be great if it makes sense. If using an artifact, consider what kind of artifact (interactive, visual, checklist, etc.) might be most helpful for this specific task. Thanks for your help! - Write video scripts [](#) this is a test [Anthropic](https://www.anthropic.com/) Anthropic © \[year\] Anthropic PBC Products - Claude [Claude](/product/overview) Claude - Claude Code [Claude Code](/product/claude-code) Claude Code - Claude Code for Enterprise [Claude Code for Enterprise](/product/claude-code/enterprise) Claude Code for Enterprise - Claude Cowork [Claude Cowork](/product/cowork) Claude Cowork - Max plan [Max plan](/pricing/max) Max plan - Team plan [Team plan](/pricing/team) Team plan - Enterprise plan [Enterprise plan](/pricing/enterprise) Enterprise plan - Download app [Download app](/download) Download app - Pricing [Pricing](/pricing) Pricing - Log in [Log in](https://claude.ai/login) Log in Features - Claude for Chrome [Claude for Chrome](/claude-for-chrome) Claude for Chrome - Claude for Slack [Claude for Slack](/claude-for-slack) Claude for Slack - Claude for Excel [Claude for Excel](/claude-for-excel) Claude for Excel - Claude for PowerPoint [Claude for PowerPoint](/claude-for-powerpoint) Claude for PowerPoint - Skills [Skills](/skills) Skills Models - Opus [Opus](../15-Claude-AI-Features/claude-opus-4-6-anthropic.md) Opus - Sonnet [Sonnet](../15-Claude-AI-Features/claude-sonnet-4-6-anthropic.md) Sonnet - Haiku [Haiku](../15-Claude-AI-Features/claude-haiku-4-5-anthropic.md) Haiku Solutions - AI agents [AI agents](/solutions/agents) AI agents - Claude Code Security [Claude Code Security](/solutions/claude-code-security) Claude Code Security - Code modernization [Code modernization](/solutions/code-modernization) Code modernization - Coding [Coding](/solutions/coding) Coding - Customer support [Customer support](/solutions/customer-support) Customer support - Education [Education](/solutions/education) Education - Financial services [Financial services](/solutions/financial-services) Financial services - Government [Government](/solutions/government) Government - Healthcare [Healthcare](/solutions/healthcare) Healthcare - Life sciences [Life sciences](/solutions/life-sciences) Life sciences - Nonprofits [Nonprofits](/solutions/nonprofits) Nonprofits Claude Platform - Overview [Overview](/platform/api) Overview - Developer docs [Developer docs](https://platform.claude.com/docs) Developer docs - Pricing [Pricing](https://claude.com/pricing#api) Pricing - Marketplace [Marketplace](/platform/marketplace) Marketplace - Amazon Bedrock [Amazon Bedrock](/partners/amazon-bedrock) Amazon Bedrock - Google Cloud’s Vertex AI [Google Cloud’s Vertex AI](/partners/google-cloud-vertex-ai) Google Cloud’s Vertex AI - Microsoft Foundry [Microsoft Foundry](/partners/microsoft-foundry) Microsoft Foundry - Regional compliance [Regional compliance](/regional-compliance) Regional compliance - Console login [Console login](../04-API-Reference/Other/platform-claude-com.md) Console login Resources - Blog [Blog](/blog) Blog - Claude partner network [Claude partner network](/partners) Claude partner network - Community [Community](/community) Community - Connectors [Connectors](/connectors) Connectors - Courses [Courses](https://www.anthropic.com/learn) Courses - Customer stories [Customer stories](/customers) Customer stories - Engineering at Anthropic [Engineering at Anthropic](https://www.anthropic.com/engineering) Engineering at Anthropic - Events [Events](https://www.anthropic.com/events) Events - Plugins [Plugins](/plugins) Plugins - Powered by Claude [Powered by Claude](/partners/powered-by-claude) Powered by Claude - Service partners [Service partners](/partners/services) Service partners - Startups program [Startups program](/programs/startups) Startups program - Tutorials [Tutorials](/resources/tutorials) Tutorials - Use cases [Use cases](/resources/use-cases) Use cases Company - Anthropic [Anthropic](https://www.anthropic.com/) Anthropic - Careers [Careers](https://www.anthropic.com/careers) Careers - Economic Futures [Economic Futures](https://www.anthropic.com/economic-futures) Economic Futures - Research [Research](https://www.anthropic.com/research) Research - News [News](https://www.anthropic.com/news) News - Responsible Scaling Policy [Responsible Scaling Policy](../19-Reference/announcing-our-updated-responsible-scaling-policy-anthropic.md) Responsible Scaling Policy - Security and compliance [Security and compliance](https://trust.anthropic.com/) Security and compliance - Transparency [Transparency](https://anthropic.com/transparency) Transparency Help and security - Availability [Availability](https://www.anthropic.com/supported-countries) Availability - Status [Status](https://status.anthropic.com/) Status - Support center [Support center](https://support.claude.com/en/) Support center Terms and policies Privacy choices ### Cookie settings We use cookies to deliver and improve our services, analyze site usage, and if you agree, to customize or personalize your experience and market our services to you. You can read our Cookie Policy [here](https://www.anthropic.com/legal/cookies). Customize cookie settings Reject all cookies Accept all cookies ###### Necessary Enables security and basic functionality. Required ###### Analytics Enables tracking of site performance. Off ###### Marketing Enables ads personalization and tracking. Off Save preferences Privacy policy [Privacy policy](https://www.anthropic.com/legal/privacy) Privacy policy Responsible disclosure policy [Responsible disclosure policy](https://www.anthropic.com/responsible-disclosure-policy) Responsible disclosure policy Terms of service: Commercial [Terms of service: Commercial](https://www.anthropic.com/legal/commercial-terms) Terms of service: Commercial Terms of service: Consumer [Terms of service: Consumer](https://www.anthropic.com/legal/consumer-terms) Terms of service: Consumer Usage policy [Usage policy](https://www.anthropic.com/legal/aup) Usage policy [x.com](https://x.com/claudeai) x.com [LinkedIn](https://www.linkedin.com/showcase/claude/) LinkedIn [YouTube](https://www.youtube.com/@anthropic-ai) YouTube [Instagram](https://www.instagram.com/claudeai) Instagram English (US) [English (US)](/) [日本語 (Japan)](/ja-jp) [Deutsch (Germany)](/de-de) [Français (France)](/fr-fr) [한국어 (South Korea)](/ko-kr) --- ## mcp-connectors-4ffa3fcb5b.md # MCP connectors April 10, 2026 MCP connectors let Claude connect to your organization’s tools, data sources, and services. Claude can search your documents, read your email, or call external APIs on your behalf, all without leaving the chat. For general questions about connectors in Claude, see **[Use connectors to extend Claude's capabilities](https://support.claude.com/en/articles/11176164-use-connectors-to-extend-claude-s-capabilities)**. ## How MCP connectors differ in Claude for Government **Claude for Government runs the first FedRAMP High authorized MCP implementation.** The MCP framework, the infrastructure that registers, authenticates, and executes Anthropic-provided connectors, operates entirely within the Claude for Government authorized boundary. Because the authorization covers MCP as a feature, new Anthropic-provided and customer-provided connectors can be added without additional FedRAMP audits of Claude for Government itself, as long as the connectors meet your agency's data-handling requirements. **The connector catalog is curated.** Only connectors approved for the government environment appear in the Admin Settings directory, so some commercial Claude connectors are not available. **Important:** Some connectors call external APIs that transmit data outside the FedRAMP boundary. Each connector's catalog description states what data flows where. Your organization is responsible for evaluating whether that data flow is appropriate for your workloads, data classification levels, and applicable regulatory requirements. ## Browse available connectors [TABLE] Organizations can also register their own MCP servers for internal systems, custom tools, or approved third-party services. See **[Custom Connectors](https://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp)**. ## Enable a connector for your organization Connectors are enabled org-wide from Admin Settings. Once enabled, individual users in your organization can connect their own accounts. **Note**: You must be an Owner or Primary Owner to enable connectors. Members can connect their accounts to connectors that are already enabled but cannot add new ones to the catalog. 1. Navigate to **claude.fedstart.com/admin-settings/connectors** (or use your organization’s custom Claude for Government domain). 2. Click "Browse connectors" to view the catalog. 3. Select a connector and review its description, paying particular attention to what data leaves the boundary, if any. 4. Click "Add to your organization." 5. Some connectors require an organization-level authentication step (for example, the Microsoft 365 connector requires tenant-admin consent in Microsoft Entra). Follow the on-screen instructions. ## Connect your account to a connector After your org admin enables a connector, it appears in your personal settings. 1. Navigate to **claude.fedstart.com/settings/connectors** 2. Click "Connect" next to the connector. 3. Complete the authentication flow—typically a one-time OAuth consent screen. 4. The connector is now active in your chats. Claude will use it automatically when relevant, or you can reference it directly ("search my SharePoint for…"). ------------------------------------------------------------------------ ## Security and privacy ### How authentication works Connectors use delegated, per-user OAuth wherever possible. When you connect a tool, Claude inherits your permissions in that tool and can only see and do what you can already see and do. There are no service accounts with elevated access. ### Where your credentials are stored OAuth tokens are encrypted at rest and stored entirely within the Claude for Government FedRAMP High boundary. Tokens are scoped to the individual user who authenticated and are never shared between users. Anthropic does not receive standing access to your connected tenants. ### What stays inside the boundary The MCPservice, token storage, and all connector-execution audit logging stay inside the FedRAMP High authorized environment. Each tool invocation follows the path: User → Claude (in boundary) → MCP service(in boundary) → connected service Whether the **connected service** sits inside or outside the boundary depends on the connector. The catalog description for each connector states this explicitly. For example, the Microsoft 365 connector calls Microsoft Graph within your tenant; the Web Search connector calls a third-party search API outside the boundary under a zero-data-retention agreement. **Important:** Adding a connector does not change Claude for Government's FedRAMP authorization, but your agency's own ATO process may require you to evaluate the specific data-handling characteristics of each connector before enabling it. ------------------------------------------------------------------------ ## Frequently asked questions ### Does enabling a connector give Anthropic access to our data? No. Connectors use per-user delegated OAuth. Data flows between the in-boundary MCP proxy and the connected service using the individual user's credentials. Anthropic does not receive standing access to your tenant. ### If we add a new connector, do we need to update our ATO? Adding a connector does not change Claude for Government's FedRAMP authorization—MCP was authorized as a feature. Whether your agency's own ATO requires an update depends on the data the connector handles and your internal authorization process. ### Can we restrict which users in our org can use a connector? Connectors are enabled at the organization level. Once enabled, any user in the org can connect their individual account. Per-user or per-group connector gating is not currently available. ### Can Claude take write actions—send emails, create files—through connectors? It depends on the connector: admins should review all connector permissions before adding them. Custom connectors can expose write tools if your MCP server implements them; see the security guidance in Custom Connectors before enabling write scopes. --- ## mcp-individual-connectors-c17c49a15d.md # MCP: Individual connectors April 9, 2026 Your organization can register its own MCP servers in Claude for Government, letting Claude connect to internal systems, custom tools, or third-party services you've approved for your environment. **Custom connectors work the same way in Claude for Government as in Claude Enterprise.** The prerequisites, the registration flow, and the behavior once enabled are identical. For full setup instructions, see **[Use connectors to extend Claude's capabilities](https://support.claude.com/en/articles/11176164-use-connectors-to-extend-claude-s-capabilities)**. This page covers only what's different. For general guidance on building and deploying MCP servers, see **[modelcontextprotocol.io](https://modelcontextprotocol.io/)** and the **[commercial connector guide](https://support.claude.com/en/articles/11176164-use-connectors-to-extend-claude-s-capabilities)**. ## How custom connectors differ for Claude for Government **Adding a connector does not trigger a new FedRAMP review** MCP was authorized as a **feature** in Claude for Government's FedRAMP High package. The infrastructure that registers, authenticates, and executes your connector—including encrypted token storage, per-user OAuth, and audit logging—is already covered. You can add new custom connectors without re-engaging Anthropic's FedRAMP audit process. **Important:** This applies to Claude for Government's authorization only. Your agency's own ATO process may still require you to evaluate the MCP server and any backing services it reaches before enabling them. Anthropic does not review, authorize, or assume responsibility for customer-registered MCP servers or the external systems they call. ## Register custom connectors Navigate to **claude.fedstart.com/admin-settings/connectors** as an **Owner** or **Primary Owner**. The rest of the flow matches the **[commercial guide](https://support.claude.com/en/articles/11176164-use-connectors-to-extend-claude-s-capabilities)**. Once added, the connector appears in your users' settings exactly as Anthropic-provided connectors do. ## Know where the FedRAMP boundary sits When a user invokes a tool from your custom connector: User → Claude → MCP proxy → Your MCP server →  Your backing service └────     inside FedRAMP High      ────┘   └─your responsibility─┘ Claude, the MCP proxy, and all OAuth token storage stay inside the FedRAMP High authorized environment. Whether your MCP server and its backing service sit inside or outside that boundary is under your control and yours to evaluate against the data classifications your users will handle. **Prefer FedRAMP-authorized backing services.** If your connector talks to a SaaS product, check whether that product has a government-cloud offering—Atlassian Government Cloud, Salesforce Government Cloud, ServiceNow GCC—and point the connector there rather than at the commercial endpoint. --- ## microsoft-365-connector-security-guide-bd172675ff.md # Microsoft 365 connector security guide The Microsoft 365 connector is an **Anthropic-hosted integration** that enables Claude to securely access Microsoft 365 services (Outlook, SharePoint, OneDrive, Teams) through user-delegated permissions. Anthropic has completed Microsoft's publisher verification process, associating our verified Microsoft Partner Network account with this application to confirm our organizational identity. The Microsoft 365 connector is available on all Claude plans: Free, Pro, Max, Team, and Enterprise. The connector operates as a **secure proxy**, and your Microsoft 365 documents, emails, and files remain in your tenant. The connector only retrieves data on-demand during active queries and doesn’t cache file content. Credentials are encrypted and managed by Anthropic's backend infrastructure. The MCP server itself doesn’t store or manage these credentials. Microsoft's Azure SDK handles the On-Behalf-Of token exchange and caching on a per-user basis for accessing the Graph API. ## Access restriction ### Access can be fully restricted The connector provides **multiple layers of access control** to address your security requirements. For detailed information on administration of the Microsoft 365 connector, see **[Set up the Microsoft 365 connector](https://support.claude.com/en/articles/12542951-)**. **1. Microsoft Entra tenant requirement** All people using the connector—regardless of Claude plan—must authenticate with a Microsoft 365 account tied to a Microsoft Entra tenant. Personal Microsoft accounts (@outlook.com, @hotmail.com) can't be used. A Microsoft Entra Global Administrator must complete a one-time consent process before anyone in the tenant can connect. **2. Organization-level gating (Team and Enterprise plans)** On Team and Enterprise plans, access to the connector requires a two-step approval process. First, Owners must explicitly enable the Microsoft 365 connector in Claude organization settings by navigating to Organization settings \> Connectors \> Browse connectors \> Add "Microsoft 365." Until this approval is granted, team members have no access. Second, after the Owner enables the connector, a Microsoft Entra Global Administrator must complete individual authentication and grant consent on behalf of the whole organization before any team members can connect. **3. Granular permission revocation** You can selectively disable specific capabilities via Microsoft Entra Admin Center. For example: [TABLE] Changes take effect immediately for all people in your organization. People can also choose to disable capabilities during a chat by selectively toggling off the connector's tools. **4. Microsoft conditional access integration** The connector fully supports your existing Entra (Azure AD) policies: - **Multi-factor authentication (MFA)**: Enforce MFA for connector access - **Device compliance**: Require managed/compliant devices - **IP restrictions**: Limit Microsoft authentication to corporate network or VPN - **Group-based access**: Restrict to specific security groups **5. User-level permissions** - The Microsoft 365 Connector uses **[delegated permissions](https://learn.microsoft.com/en-us/graph/permissions-overview?tabs=http#delegated-permissions)**. - Users can only access Microsoft 365 data **they already have permission** for - SharePoint search requires Sites.Read.All permission. Site-specific permissioning (using \*.Selected permissions) is not supported because the underlying search is tenant-wide. - Users cannot bypass SharePoint sharing settings or folder permissions. - Users cannot access other users' private files or emails. Users can search shared mailboxes they've been granted delegate access to in Microsoft 365, including full access and folder-level delegation. Shared mailbox access remains read-only, via the `Mail.Read.Shared` permission. - Delegated permissions inherently respect Microsoft 365 data loss prevention (DLP) policies. **6. Token management** - Refresh tokens expire after 90 days of inactivity by default, requiring re-authentication. This can be customized in Microsoft Entra ID using a token lifetime policy. - Access tokens typically expire within 60-90 minutes per Microsoft Entra ID defaults and are automatically refreshed. - Admins or users can revoke access anytime via Microsoft Entra ID. - The Microsoft 365 Connector never sees or stores passwords. ## Security architecture summary ### Authentication flow - **OAuth 2.0 On-Behalf-Of (OBO):** Industry-standard delegated authentication - **PKCE protection**: Public client uses Proof Key for Code Exchange to prevent authorization code interception - **Two-stage token exchange**: User authenticates to obtain access token for MCP server, then MCP server exchanges it for Graph API access using OBO flow with confidential client credentials. In this flow, not even the user or their Claude client has access to the OBO tokens. Only the MCP server can access and use tokens with access to the user’s data via the Microsoft Graph API. - **No credential storage**: Users never share Microsoft passwords with Anthropic - **Encrypted token storage**: Access and refresh tokens are encrypted while cached by the Claude backend ### Data flow - Documents and other content are retrieved **only during active queries** - **Tool call results** from the connector that are **part of stored chats are retained** - **The user who requested the Claude chat** can see the tool call results and Claude’s response incorporating the data - **Other users shared on the chat** can only see Claude’s response incorporating the result of the tool call - Each request creates a fresh data flow which is cleaned up after the response is returned ### Multi-tenant isolation - Microsoft Entra tenants are **cryptographically separated** from each other using a common-scoped multi-tenant configuration - Multi-tenant isolation is cryptographically enforced through digitally signed access tokens that bind each user to their organization’s tenant ## Available capabilities ### Read and search tools The connector provides **read-only** access to: [TABLE] ### Write tools [TABLE] **Note:** “Always allow” is not supported for `outlook_send_email`, `outlook_forward_mail`, `outlook_send_draft`, `outlook_create_event`, or `outlook_update_event`. When an organization enables write tools, the connector also exposes write tools for sending and organizing email, managing drafts and calendar events, updating mailbox settings, and creating and updating files in OneDrive and SharePoint. Teams remains read-only. Write tools include the following built-in safeguards: - **Attribution:** Emails Claude sends include an attribution header identifying them as agent-initiated. File and calendar writes aren't currently tagged. - **Rate limits:** Per-user limits apply to writes, sends, and recipients. - **Attachment restriction:** Attachments aren't supported in any write tool—sending, forwarding, and drafting all reject messages with attachments. - **Blocked by default:** Organizations that used the connector before write tools launched have write tools blocked by default until an admin enables them. ## Permissions list **Basic permissions** - **[User.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#userread)** - Sign in and read user profile (basic requirement) **Mail permissions** - **[Mail.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#mailread)** - Read user mail (required for email tools/resources) - **[Mail.ReadBasic](https://learn.microsoft.com/en-us/graph/permissions-reference#mailreadbasic)** - Read user mail metadata (alternative for limited functionality) - **[Mail.Read.Shared](https://learn.microsoft.com/en-us/graph/permissions-reference#mailreadshared)** - Read user and shared mail - **[MailboxFolder.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#mail-permissions)** - Read a user's mailbox folders - **[MailboxItem.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#mail-permissions)** - Read a user's mailbox items **Calendar permissions** - **[Calendars.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#calendarsread)** - Read user calendars and events - **[Calendars.Read.Shared](https://learn.microsoft.com/en-us/graph/permissions-reference#calendarsreadshared)** - Read calendars user can access, including shared **User directory** - **[User.ReadBasic.All](https://learn.microsoft.com/en-us/graph/permissions-reference#userreadbasicall)** - Read basic profiles of all users (for meeting availability) **Chat permissions** - **[Chat.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#chatread)** - Read user chat messages - **[Chat.ReadBasic](https://learn.microsoft.com/en-us/graph/permissions-reference#chatreadbasic)** - Read user chat metadata (alternative for limited functionality) - **[ChatMember.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#chatmemberread)** - Read the members of chats - **[ChatMessage.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#chatmessageread)** - Read user chat messages (more specific than Chat.Read) **Channel permissions** - **[Channel.ReadBasic.All](https://learn.microsoft.com/en-us/graph/permissions-reference#channelreadbasicall)** - Read the names and descriptions of channels - **[ChannelMessage.Read.All](https://learn.microsoft.com/en-us/graph/permissions-reference#channelmessagereadall)** - Read channel messages **Meeting permissions** - **[OnlineMeetings.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#onlinemeetingsread)** - Read online meetings - **[OnlineMeetingTranscript.Read.All](https://learn.microsoft.com/en-us/graph/permissions-reference#onlinemeetingtranscriptreadall)** - Read meeting transcripts - **[OnlineMeetingAiInsight.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#onlinemeetingaiinsightread)** - Read all AI Insights for online meetings - **[OnlineMeetingArtifact.Read.All](https://learn.microsoft.com/en-us/graph/permissions-reference#onlinemeetingartifactreadall)** - Read user's online meeting artifacts - **[OnlineMeetingRecording.Read.All](https://learn.microsoft.com/en-us/graph/permissions-reference#onlinemeetingrecordingreadall)** - Read all recordings of online meetings **Files permissions** - **[Files.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#filesread)** - Read user files - **[Files.Read.All](https://learn.microsoft.com/en-us/graph/permissions-reference#filesreadall)** - Read all files user can access **Sites permissions** - **[Sites.Read.All](https://learn.microsoft.com/en-us/graph/permissions-reference#sitesreadall)** - Read items in all site collections **Write permissions** Requested as part of the updated consent set; used only when write tools are enabled: - **[Mail.Send](https://learn.microsoft.com/en-us/graph/permissions-reference#mailsend)** - Send and forward email - **[Mail.ReadWrite](https://learn.microsoft.com/en-us/graph/permissions-reference#mailreadwrite)** - Create, update, and delete drafts; move and label messages - **[Calendars.ReadWrite](https://learn.microsoft.com/en-us/graph/permissions-reference#calendarsreadwrite)** - Create, update, delete, and respond to calendar events - **[Files.ReadWrite.All](https://learn.microsoft.com/en-us/graph/permissions-reference#filesreadwriteall)** - Create and update files in OneDrive and SharePoint - **[MailboxSettings.ReadWrite](https://learn.microsoft.com/en-us/graph/permissions-reference#mailboxsettingsreadwrite)** - Manage categories, inbox rules, and automatic replies ## Current limitations - **Teams is read-only**: Claude can't post Teams messages or modify Teams settings. Other write tools require an admin to enable them. - **User-level access only**: Access with service principal authentication is not supported. ## Frequently asked questions ### Can we test with a small pilot group before enterprise-wide rollout? Yes. The recommended approach is to use app assignment to restrict who can use the connector: - Enable the connector (Team and Enterprise Owners enable it in organization settings; individual plan users can connect directly). - Microsoft Entra Admin completes pre-consent setup - Use Microsoft Entra Enterprise App assignment to restrict access to specific users or groups (e.g., assign only "IT Security Test Group" to the app). - Expand groups progressively for gradual deployment ### How do we ensure no data leakage occurs between our organization and others in the multi-tenant environment? Multi-tenant isolation ensures complete separation: - Server uses the common tenant configuration to accept tokens from any Microsoft Entra ID tenant - Each user's token contains their organization's tenant ID (tid claim) which is validated - Graph API tokens obtained through OBO are automatically scoped to the user and their tenant - Cross-tenant token access is prevented cryptographically by the design of Microsoft Graph’s OAuth 2.0 implementation. ### What happens if someone tries to connect with a personal Microsoft account? The connector requires a Microsoft Entra tenant tied to a Microsoft Business plan. Personal Microsoft accounts (@outlook.com, @hotmail.com) can't be used to authenticate. People attempting to connect with a personal account will receive an authentication error. ### Do you have audit logging for compliance? Yes. All Graph API calls made by the connector are logged in your organization's Microsoft 365 audit log, which you can access through the M365 Compliance Center. These logs show the timestamp, user, operation performed, and resource accessed, with retention periods matching your Microsoft 365 audit policy. Additionally, Anthropic logs authentication and tool execution events. ### Can we revoke access if we discover unauthorized usage? There are multiple revocation methods: - **Individual:** Users disconnect via **Customize \> Connectors** - **Admin-level**: On Team and Enterprise plans, Owners disable the connector in Claude organization settings (all team members affected). - **Permission-level**: Revoke specific permissions in Microsoft Entra Admin Center - **Tenant-level**: Revoke all permissions in Microsoft Entra Admin Center ### What certifications does Anthropic have? Anthropic has the following certifications: - **SOC 2 Type II** (annual audit) - **ISO 27001** certified - **GDPR compliant** (DPA available) - **Microsoft publisher-verified** application ## Additional resources - **[Set up the Microsoft 365 connector](https://support.claude.com/en/articles/12542951-)** - **[Connect Claude to Microsoft 365](https://support.claude.com/en/articles/15183774)** - **[Overview of Microsoft Graph permissions: Delegated permissions](https://learn.microsoft.com/en-us/graph/permissions-overview?tabs=http#delegated-permissions)** --- ## microsoft-365-connector-security-guide.md 4. Microsoft 365 Connector: Security Guide # Microsoft 365 Connector: Security Guide ## What it is The Microsoft 365 Connector is an **Anthropic-hosted integration** that enables Claude to securely access Microsoft 365 services (Outlook, SharePoint, OneDrive, Teams) through user-delegated permissions. Anthropic has completed Microsoft's publisher verification process, associating our verified Microsoft Partner Network account with this application to confirm our organizational identity. The connector operates as a **secure proxy**, and your Microsoft 365 documents, emails, and files remain in your tenant. The connector only retrieves data on-demand during active queries and does not cache file content. Credentials are encrypted and managed by Anthropic's backend infrastructure. The MCP server itself does not store or manage these credentials. Microsoft's Azure SDK handles the On-Behalf-Of token exchange and caching on a per-user basis for accessing the Graph API. ## Access restriction ### Access can be fully restricted The connector provides **multiple layers of access control** to address your security requirements. For detailed information on administration of the Microsoft 365 Connector see **[Enabling and using the Microsoft 365 connector](enabling-and-using-the-microsoft-365-connector.md)**. **1. Organization-level gating** Access to the connector for Team and Enterprise plan users requires a two-step approval process. First, owners must explicitly enable the Microsoft 365 connector in Claude organization settings by navigating to Organization settings → Connectors → Browse connectors → Add "Microsoft 365". Until this approval is granted, users have no access. Second, after the Owner enables the connector, a Microsoft Entra Global Administrator must complete individual authentication and grant consent on behalf of the whole organization before any team members can connect. **2. Microsoft Entra Admin pre-consent requirement** Before users can access the connector, a Microsoft Entra Admin must complete a one-time setup, which will: - Add two service principals and Enterprise apps in Microsoft Entra ID (M365 MCP Client and M365 MCP Server). This establishes a service-level identity for the Microsoft 365 Connector apps in your tenant. - Grant admin pre-consent for your Microsoft 365 tenant. - Optionally restrict which Microsoft Entra ID users and groups are allowed to use the connector. - Optionally restrict permissions the connector is allowed to use to selectively control which Microsoft 365 services are accessible. **3. Granular permission revocation** You can selectively disable specific capabilities via Microsoft Entra Admin Center. For example: [TABLE] Changes take effect immediately for all users in your organization. Note that users can also choose to disable capabilities that they have permission to use during a chat or session by selectively toggling off the connector’s tools. **4. Microsoft conditional access integration** The connector fully supports your existing Entra (Azure AD) policies: - **Multi-factor authentication (MFA)**: Enforce MFA for connector access - **Device compliance**: Require managed/compliant devices - **IP restrictions**: Limit Microsoft authentication to corporate network or VPN - **Group-based access**: Restrict to specific security groups **5. User-level permissions** - The Microsoft 365 Connector uses **[delegated permissions](https://learn.microsoft.com/en-us/graph/permissions-overview?tabs=http#delegated-permissions)**. - Users can only access Microsoft 365 data **they already have permission** for - SharePoint search requires Sites.Read.All permission. Site-specific permissioning (using \*.Selected permissions) is not supported because the underlying search is tenant-wide. - Users cannot bypass SharePoint sharing settings or folder permissions - Users cannot access other users' private files or emails - Delegated permissions inherently respect Microsoft 365 data loss prevention (DLP) policies **6. Token management** - Refresh tokens expire after 90 days of inactivity by default, requiring re-authentication. This can be customized in Microsoft Entra ID using a token lifetime policy. - Access tokens typically expire within 60-90 minutes per Microsoft Entra ID defaults and are automatically refreshed - Admins or users can revoke access anytime via Microsoft Entra ID - The Microsoft 365 Connector never sees or stores passwords ## Security architecture summary ### Authentication flow - **OAuth 2.0 On-Behalf-Of (OBO):** Industry-standard delegated authentication - **PKCE protection**: Public client uses Proof Key for Code Exchange to prevent authorization code interception - **Two-stage token exchange**: User authenticates to obtain access token for MCP server, then MCP server exchanges it for Graph API access using OBO flow with confidential client credentials. In this flow, not even the user or their Claude client has access to the OBO tokens. Only the MCP server can access and use tokens with access to the user’s data via the Microsoft Graph API. - **No credential storage**: Users never share Microsoft passwords with Anthropic - **Encrypted token storage**: Access and refresh tokens are encrypted while cached by the Claude backend ### Data flow - Documents and other content are retrieved **only during active queries** - **Tool call results** from the connector that are **part of stored chats are retained** - **The user who requested the Claude chat** can see the tool call results and Claude’s response incorporating the data - **Other users shared on the chat** can only see Claude’s response incorporating the result of the tool call - Each request creates a fresh data flow which is cleaned up after the response is returned ### Multi-tenant isolation - Microsoft Entra tenants are **cryptographically separated** from each other using a common-scoped multi-tenant configuration - Multi-tenant isolation is cryptographically enforced through digitally signed access tokens that bind each user to their organization’s tenant ## Available capabilities ### Current features (read-only access) The connector provides **read-only** access to: [TABLE] ## Permissions list **Basic permissions** - [User.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#userread) - Sign in and read user profile (basic requirement) **Mail permissions** - [Mail.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#mailread) - Read user mail (required for email tools/resources) - [Mail.ReadBasic](https://learn.microsoft.com/en-us/graph/permissions-reference#mailreadbasic) - Read user mail metadata (alternative for limited functionality) - [Mail.Read.Shared](https://learn.microsoft.com/en-us/graph/permissions-reference#mailreadshared) - Read user and shared mail - [MailboxFolder.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#mail-permissions) - Read a user's mailbox folders - [MailboxItem.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#mail-permissions) - Read a user's mailbox items **Calendar permissions** - [Calendars.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#calendarsread) - Read user calendars and events - [Calendars.Read.Shared](https://learn.microsoft.com/en-us/graph/permissions-reference#calendarsreadshared) - Read calendars user can access, including shared **User directory** - [User.ReadBasic.All](https://learn.microsoft.com/en-us/graph/permissions-reference#userreadbasicall) - Read basic profiles of all users (for meeting availability) **Chat permissions** - [Chat.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#chatread) - Read user chat messages - [Chat.ReadBasic](https://learn.microsoft.com/en-us/graph/permissions-reference#chatreadbasic) - Read user chat metadata (alternative for limited functionality) - [ChatMember.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#chatmemberread) - Read the members of chats - [ChatMessage.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#chatmessageread) - Read user chat messages (more specific than Chat.Read) **Channel permissions** - [Channel.ReadBasic.All](https://learn.microsoft.com/en-us/graph/permissions-reference#channelreadbasicall) - Read the names and descriptions of channels - [ChannelMessage.Read.All](https://learn.microsoft.com/en-us/graph/permissions-reference#channelmessagereadall) - Read channel messages **Meeting permissions** - [OnlineMeetings.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#onlinemeetingsread) - Read online meetings - [OnlineMeetingTranscript.Read.All](https://learn.microsoft.com/en-us/graph/permissions-reference#onlinemeetingtranscriptreadall) - Read meeting transcripts - [OnlineMeetingAiInsight.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#onlinemeetingaiinsightread) - Read all AI Insights for online meetings - [OnlineMeetingArtifact.Read.All](https://learn.microsoft.com/en-us/graph/permissions-reference#onlinemeetingartifactreadall) - Read user's online meeting artifacts - [OnlineMeetingRecording.Read.All](https://learn.microsoft.com/en-us/graph/permissions-reference#onlinemeetingrecordingreadall) - Read all recordings of online meetings **Files permissions** - [Files.Read](https://learn.microsoft.com/en-us/graph/permissions-reference#filesread) - Read user files - [Files.Read.All](https://learn.microsoft.com/en-us/graph/permissions-reference#filesreadall) - Read all files user can access **Sites permissions** - [Sites.Read.All](https://learn.microsoft.com/en-us/graph/permissions-reference#sitesreadall) - Read items in all site collections - [Sites.Selected](https://learn.microsoft.com/en-us/graph/permissions-reference#sitesselected) – Manages application access at the site collection level, providing access to a specific site collection ## Current limitations - **No write capabilities**: Cannot send emails, schedule meetings, create/modify documents, or post Teams messages. - **User-level access only**: Access with service principal authentication is not supported. ## Frequently asked questions ### Q: Can we test with a small pilot group before enterprise-wide rollout? A: Yes. The recommended approach is to use app assignment to restrict who can use the connector: - Owner enables connector in Claude organization settings - Microsoft Entra Admin completes pre-consent setup - Use Microsoft Entra Enterprise App assignment to restrict access to specific users or groups (e.g., assign only "IT Security Test Group" to the app). - Expand groups progressively for gradual deployment ### Q: How do we ensure no data leakage occurs between our organization and others in the multi-tenant environment? **A**: Multi-tenant isolation ensures complete separation: - Server uses the common tenant configuration to accept tokens from any Microsoft Entra ID tenant - Each user's token contains their organization's tenant ID (tid claim) which is validated - Graph API tokens obtained through OBO are automatically scoped to the user and their tenant - Cross-tenant token access is prevented cryptographically by the design of Microsoft Graph’s OAuth 2.0 implementation. ### Q: What happens if an employee tries to access company data from a personal Claude account? **A**: The connector validates identity during authentication: - User must have access to the Team/Enterprise organization where the connector is enabled - Microsoft login validates user's Microsoft Entra ID credentials - Token validation confirms user's tenant ID - Graph API enforces tenant boundaries - Personal Claude accounts cannot access enterprise Microsoft 365 data without organization membership ### Q: Do you have audit logging for compliance? **A:** Yes, audit logging is available for your compliance needs. All Graph API calls made by the connector are logged in your organization's Microsoft 365 audit log, which you can access through the M365 Compliance Center. These logs show the timestamp, user, operation performed, and resource accessed, with retention periods matching your Microsoft 365 audit policy. Additionally, Anthropic logs authentication and tool execution events. ### Q: Can we revoke access if we discover unauthorized usage? **A**: There are multiple revocation methods: - **User-level**: Users disconnect via Claude Settings → Connectors - **Admin-level**: Disable connector in Claude organization settings (all users affected) - **Permission-level**: Revoke specific permissions in Microsoft Entra Admin Center - **Tenant-level**: Revoke all permissions in Microsoft Entra Admin Center ### Q: What certifications does Anthropic have? **A:** Anthropic has the following certifications: - **SOC 2 Type II** (annual audit) - **ISO 27001** certified - **GDPR compliant** (DPA available) - **Microsoft publisher-verified** application ## Additional resources - Claude Help Center: **[Enabling and using the Microsoft 365 connector](enabling-and-using-the-microsoft-365-connector.md)** - **[Overview of Microsoft Graph permissions: Delegated permissions](https://learn.microsoft.com/en-us/graph/permissions-overview?tabs=http#delegated-permissions)** --- ## pre-built-web-connectors-using-remote-mcp-baa2f1264a.md # Use connectors to extend Claude's capabilities This guide explains how to enable and use connectors with Claude to enhance its capabilities. Web connectors are available for all users on Claude, Cowork, Claude Desktop, and Claude Mobile (iOS and Android). Desktop extensions are available to all users on Claude Desktop. ## What are connectors? Connectors let Claude access your apps and services, retrieve your data, and take actions within connected services. Claude inherits each person's permissions from the connected service. If someone can't access a specific file, channel, or record in the source system, the connector can't reach it from Claude either. For example, you can connect Claude to Linear to create issues, to Slack to send messages, or to Google Drive to search your files. Connectors work across Claude, Claude Desktop, Claude Code, and the API (via the **[MCP Connector](https://platform.claude.com/docs/en/agents-and-tools/mcp-connector)**). You can find available connectors in the **[Connectors Directory](https://claude.ai/connectors)**, where each connector has a page detailing its use cases, read/write capabilities, and availability. You can also add custom connectors or connect to any service that supports MCP. ## How Claude suggests connected apps Once you've connected an app, Claude can bring it into a conversation on its own when it fits what you're asking for—you don't have to name it every time. For details on how that works, including what happens when more than one app could help, see **[How Claude suggests connected apps](https://support.claude.com/en/articles/14730684-how-claude-suggests-connected-apps)**. ## Browse available connectors You can browse the Connectors Directory on Claude, Claude Desktop, and Claude for iOS/Android. You can browse the directory from two areas: **From a chat** 1. Click the "+" button in the lower left corner of your chat, or type "/" to open the menu. 2. Hover over “Connectors.” 3. Select “Manage connectors.” 4. Click the “+” button next to **Connectors**. 5. This opens a **Connectors** modal where you can see available connectors by category or scroll through the complete list. **From settings** 1. Navigate to **[Customize \> Connectors](http://claude.ai/customize/connectors)**. 2. Click the “+” button next to **Connectors**. 3. Browse available connectors by category or scroll through the complete list. ------------------------------------------------------------------------ ## Connect a service to Claude You can connect Claude to a new service on Claude, Claude Desktop, and Claude for iOS and Android. Installing connectors on mobile is currently in beta—Claude Desktop and web remain the primary path for custom connectors. To connect a service from the directory: 1. Click on the connector you want to add. 2. Review the connector's description and capabilities. 3. Click “Connect” or “Install” to begin the setup process. 4. Follow the authentication prompts to grant Claude access to your account. 5. Configure any specific settings or permissions as needed. **Important:** When connecting to a service, you're granting Claude permission to access and potentially modify data within that service based on your account permissions. Only connect services you trust and need for your workflows. ## Connect a service on Team and Enterprise plans Before members of Team and Enterprise plans can use connectors, an Owner or Primary Owner needs to enable them for the organization. Enabling a connector makes it available to your team, but it doesn't automatically grant anyone access. Each person still needs to authenticate individually before they can use it. **Note:** With Enterprise-managed auth, you authorize a connector once for your entire organization, and your team inherits access automatically on first login. Enterprise-managed auth is available in beta for Team and Enterprise plans on Claude. Learn about **[authorizing MCP connectors for your entire organization](https://support.claude.com/en/articles/15537633)**. 1. Navigate to **[Organization settings \> Connectors](https://claude.ai/admin-settings/connectors)**. 2. Click “Browse connectors” at the bottom of the page. 3. Select the connector from the list and click “Add to your team.” 4. Individual users can then authenticate with the connector to start using it with Claude. Once enabled at the org level, individual users follow the same steps described above to connect and authenticate. **Note:** On Enterprise plans, Owners and Primary Owners can prevent services on their verified domains from being connected to Claude accounts outside the organization. Learn more about **[restricting verified-domain connectors to your enterprise](https://support.claude.com/en/articles/15402193-restrict-verified-domain-connectors-to-your-enterprise)**. ### Restrict actions within connected services Owners on Team and Enterprise plans can limit which actions a connected service can take across your organization. For example, you can allow a connector to read data from a service while preventing it from writing any changes back. This applies org-wide to everyone using the connector—individual users can't override it. Common use cases: - Allow Claude to search and summarize email, but prevent it from sending messages. - Allow Claude to read files in Google Drive, but prevent it from creating or editing documents. - Allow Claude to view Linear issues, but prevent it from creating new ones or changing status. To configure action restrictions: 1. Navigate to **[Customize \> Connectors](http://claude.ai/customize/connectors).** 2. Select the connector to see **Tool permissions**. 3. The permissions will be categorized by type (for example, read-only tools, write/delete tools). 4. For each permission category or individual permission, select Always allow, Needs approval, or Blocked. **Note:** Action restrictions work alongside source-system permissions. Even when you allow a write action in Claude, a person still needs the underlying permission in the source system to make that change. Restricting actions in Claude never grants more access than the source system permits—it only narrows it. ------------------------------------------------------------------------ ## Use connected services Once you connect to a service on Claude or Claude Desktop, it will be available to use the next time you log in to your account on Claude for iOS or Android. Once connected, services become available in your conversations: 1. Click the “+” button in the lower left of the chat interface (you can also type “/” to open this menu). 2. Hover over “Connectors” to open the menu. 3. Enable the specific services you want Claude to use for that conversation by toggling them on. 4. Claude will now be able to use these services when relevant to your requests. For example, after connecting Linear, you can ask Claude to "Create a new issue for the login bug" and Claude will use Linear to create the issue in your workspace. Some connectors are interactive and can render live interfaces—like dashboards, task boards, and design tools—directly within your conversation. Look for the **Interactive** badge in the Connectors Directory to find connectors with this capability. ------------------------------------------------------------------------ ## Choose how connectors load in your conversation When you add many connectors, Claude gives you control over how they load. You can find the **Tool access** setting by clicking the “+” button in the lower left corner of your chat, or typing “/” to open the menu. Hover over “Connectors,” then “Tool access” to choose your preferred mode. For most users, **Auto** (the default) works well. If you have 10 or more connectors active, consider switching to **On demand** to give your conversations more room. Learn more about tool access modes: **[Manage Claude's tool access](https://support.claude.com/en/articles/13730515-manage-claude-s-tool-access)**. ------------------------------------------------------------------------ ## Manage your connectors To manage your connected services: 1. Navigate to **[Customize \> Connectors](http://claude.ai/customize/connectors)**. 2. View all your connected services in the **Connectors** section. 3. For each service, you can disconnect it, modify connection settings, or review permissions and access levels. ------------------------------------------------------------------------ ## Custom connectors Custom connectors using remote MCP are available on Claude, Cowork, and Claude Desktop for users on free, Pro, Max, Team, and Enterprise plans. Free users are limited to one custom connector. In addition to directory connectors, you can add custom connectors: 1. Navigate to **[Customize \> Connectors](http://claude.ai/customize/connectors)**. 2. Click the “+” button next to **Connectors**. 3. Select “Add custom connector.” 4. Enter the connector's name and URL. 5. Enter advanced settings (OAuth Client ID and secret) if desired. 6. Click “Add,” then follow the same connection process as directory connectors. Custom connectors connect to your MCP server from Anthropic's cloud, not from your local device. Your server must be reachable over the public internet. If it's behind a firewall or on a private network, see **[Get started with custom connectors using remote MCP](https://support.claude.com/en/articles/11175166)** for network requirements and private network options. **Important:** Custom connectors allow you to connect Claude to services that haven't been verified by Anthropic. Only connect to servers from trusted organizations and review authentication permissions carefully. ------------------------------------------------------------------------ ## Security and privacy All data transfers are encrypted. When using connectors, you can only sync content to Claude that you have permission to view in the original source. Connected services process data on their own infrastructure, under their own terms, which may be located outside the United States. Settings that control where Claude's inference runs, like the US-only inference setting on Enterprise plans, don't change where third-party services operate. When connecting to services from the directory, review what access the service is requesting during the connection process. Disconnect services you no longer need or use. **For Team and Enterprise plans:** - Access permissions are enforced at the user level. Users need to authenticate with the third-party service when first using the connector, even after an Owner or Primary Owner enables it. - Connectors are only available in private projects. - Chats with synced content can't be shared. ------------------------------------------------------------------------ ## Troubleshoot connection issues If you're having trouble connecting to a service, try these steps: 1. Check that you have a stable internet connection. 2. Verify you have an active account with the service. 3. Review any permissions or account type requirements for the service. 4. If authentication fails, try disconnecting and reconnecting from **[Customize \> Connectors](http://claude.ai/customize/connectors)**. ### See a message that says, "This corporate identity belongs to an Enterprise that manages access through their own Claude account"? The service you're trying to connect uses an email address on a domain that an Enterprise organization has verified, and that organization restricts connections to its own Claude accounts only. To use this connection, sign in to your organization's Claude account and connect the service there. If you don't have a Claude account in that organization, contact your admin for access. Admins can learn more about **[restricting verified-domain connectors to your Enterprise](https://support.claude.com/en/articles/15402193-restrict-verified-domain-connectors-to-your-enterprise)**. ### Custom connector not connecting or timing out? Custom connectors (remote MCP servers) are reached from Anthropic's cloud infrastructure, not from your local machine. This is true even if you're using Cowork or Claude Desktop, which run locally on your computer. If your MCP server is behind a corporate firewall, on a private network, or not reachable over the public internet, the connection will fail. To resolve this, you can either allowlist Anthropic's IP ranges in your firewall to create a secure outbound-only connection from your network. For detailed guidance, see the network requirements section in **[Get started with custom connectors using remote MCP](https://support.claude.com/en/articles/11175166)**. --- ## resources-tutorials-how-to-use-claude-in-excel-for-accounting-revenue-model-vali.md :root { --grid-breakout: [full-start] minmax(0, 1fr) [content-start] repeat(var(--_grid---column-count), minmax(0, var(--_grid---column-width))) [content-end] minmax(0, 1fr) [full-end]; --grid-breakout-single: [full-start] minmax(0, 1fr) [content-start] minmax(0, calc(100% - var(--site--margin) * 2)) [content-end] minmax(0, 1fr) [full-end]; } ::before, ::after { box-sizing: border-box; } .w-embed:before, .w-embed:after, .w-richtext:before, .w-richtext:after { content: unset; } html { background-color: var(--_theme---background); } button { background-color: unset; padding: unset; text-align: inherit; } button:not(:disabled) { cursor: pointer; } video { width: 100%; object-fit: cover; } /* remove padding of empty element */ .wf-empty { padding: 0; } svg { max-width: 100%; } @media (prefers-color-scheme: light) { option { color: black; } } img::selection { background: transparent; } /* Typography */ body { text-transform: var(--_text-style---text-transform); font-smoothing: antialiased; -webkit-font-smoothing: antialiased; } /* Clear Defaults */ a:not ([class]) { text-decoration: underline;} [class~="u-rich-text"] a, [class~="u-rich-text-cs"] a, [class~="u-rich-text-blog"] a, [class~="u-rich-text-tutorials"] a, a.u-rich-text, [class~="command_instruction"] a { transition: color .15s ease-out, text-decoration-color .15s ease-out; text-underline-offset: 3px; text-decoration: underline; color: currentcolor; text-decoration-color: var(--_theme---border-primary); } [class~="u-rich-text"] a:hover, [class~="u-rich-text-cs"] a:hover, [class~="u-rich-text-blog"] a:hover, [class~="u-rich-text-tutorials"] a:hover, a.u-rich-text:hover, [class~="command_instruction"] a:hover { text-decoration-color: var(--_theme---foreground-primary); color: var(--_theme---foreground-primary); } h1,h2,h3,h4,h5,h6,p,blockquote,label { font-family: inherit; font-size: inherit; font-weight: inherit; line-height: inherit; letter-spacing: inherit; text-transform: inherit; text-wrap: inherit; margin-top: 0; margin-bottom: 0; } select:has(option[value=""]:checked) { color: color-mix(in lab, currentcolor 60%, transparent) } /* Selection Color */ ::selection { background-color: var(--_theme---selection--background); color: var(--_theme---selection--text); } /* Margin Trim */ :is(.u-margin-trim,.u-rich-text) > :not(:not(.w-condition-invisible,.u-cover-absolute,.u-ignore-trim) ~ :not(.w-condition-invisible,.u-cover-absolute,.u-ignore-trim)), :is(.u-margin-trim,.u-rich-text) > :not(:not(.w-condition-invisible,.u-cover-absolute,.u-ignore-trim) ~ :not(.w-condition-invisible,.u-cover-absolute,.u-ignore-trim)).u-display-contents > :first-child { margin-top: 0; } :is(.u-margin-trim,.u-rich-text) > :not(:has(~ :not(.w-condition-invisible,.u-cover-absolute,.u-ignore-trim))), :is(.u-margin-trim,.u-rich-text) > :not(:has(~ :not(.w-condition-invisible,.u-cover-absolute,.u-ignore-trim))).u-display-contents > :last-child { margin-bottom: 0; } /* Line Height Trim */ :is(h1,h2,h3,h4,h5,h6,p):not(.u-text-trim-off,:has([class*="u-text-style-"]))::before, [class*="u-text-style-"]:not(.u-text-trim-off,:has(h1,h2,h3,h4,h5,h6,p))::before { content: ""; display: table; margin-bottom: calc(-0.5lh + var(--_text-style---trim-top)); } :is(h1,h2,h3,h4,h5,h6,p):not(.u-text-trim-off,:has([class*="u-text-style-"]))::after, [class*="u-text-style-"]:not(.u-text-trim-off,:has(h1,h2,h3,h4,h5,h6,p))::after { content: ""; display: table; margin-bottom: calc(-0.5lh + var(--_text-style---trim-bottom)); } /* Rich Text Links */ .w-richtext a { position: relative; z-index: 4; } /* Line Clamp */ .u-line-clamp-1, .u-line-clamp-2, .u-line-clamp-3, .u-line-clamp-4 { -webkit-line-clamp: 1; -webkit-box-orient: vertical; } .u-line-clamp-2 { -webkit-line-clamp: 2; } .u-line-clamp-3 { -webkit-line-clamp: 3; } .u-line-clamp-4 { -webkit-line-clamp: 4; } /* Child Contain */ .u-child-contain > * { width: 100%; max-width: inherit !important; margin-inline: 0 !important; margin-top: 0 !important; } /* Hide */ .u-hide-if-empty:empty, .u-hide-if-empty:not(:has(> :not(.w-condition-invisible))), .u-hide-if-empty-cms:not(:has(.w-dyn-item)), .u-embed-js, .u-embed-css { display: none !important; } /* Focus State */ a, button, :where([tabindex]), [data-outline] { outline-offset: var(--focus--offset-outer); } a:focus-visible, button:focus-visible, [tabindex]:focus-visible, label:has(input:focus-visible) [data-outline] { outline-color: color-mix(in srgb, var(--_button-style---border) 50%, transparent); outline-width: var(--focus--width); outline-style: solid; } /* Global / Clickable Component */ .wf-design-mode .clickable_wrap { z-index: 0; } .clickable_wrap a[href="#"] { display: none; } .clickable_wrap a[href="#"] ~ button { display: block; } /* Responsive Above */ @container threshold-large (width >= 62em) { .u-order-unset-above { order: unset !important; } .u-all-unset-above { all: unset !important; } .u-grid-below { display: flex !important; } .u-max-width-unset-above { max-width: unset !important; } .u-width-unset-above { width: unset !important; } .u-hide-above { display: none !important; } } @container threshold-medium (width >= 48em) { .u-order-unset-above { order: unset !important; } .u-all-unset-above { all: unset !important; } .u-grid-below { display: flex !important; } .u-max-width-unset-above { max-width: unset !important; } .u-width-unset-above { width: unset !important; } .u-hide-above { display: none !important; } } @container threshold-small (width >= 30em) { .u-order-unset-above { order: unset !important; } .u-all-unset-above { all: unset !important; } .u-grid-below { display: flex !important; } .u-max-width-unset-above { max-width: unset !important; } .u-width-unset-above { width: unset !important; } .u-hide-above { display: none !important; } } /* Responsive Below */ @container threshold-large (width < 62em) { .u-order-unset-below { order: unset !important; } .u-all-unset-below { all: unset !important; } .u-grid-above { display: flex !important; } .u-max-width-unset-below { max-width: unset !important; } .u-width-unset-below { width: unset !important; } .u-alignment-unset-below { --_alignment---direction: start; align-self: start; } .u-hide-below { display: none !important; } } @container threshold-medium (width < 48em) { .u-order-unset-below { order: unset !important; } .u-all-unset-below { all: unset !important; } .u-grid-above { display: flex !important; } .u-max-width-unset-below { max-width: unset !important; } .u-width-unset-below { width: unset !important; } .u-alignment-unset-below { --_alignment---direction: start; align-self: start; } .u-hide-below { display: none !important; } } @container threshold-small (width < 30em) { .u-order-unset-below { order: unset !important; } .u-all-unset-below { all: unset !important; } .u-grid-above { display: flex !important; } .u-max-width-unset-below { max-width: unset !important; } .u-width-unset-below { width: unset !important; } .u-alignment-unset-below { --_alignment---direction: start; align-self: start; } .u-hide-below { display: none !important; } } /* Form Radio */ .form_main_radio_label:has(input:checked) .form_main_radio_circle_inner { opacity: 1; } /* Form Checkbox */ .form_main_checkbox_label:has(input:checked) .form_main_checkbox_box { background-color: currentColor; border-color: currentColor; } .form_main_checkbox_label:has(input:checked) .form_main_checkbox_icon { opacity: 1; } /* State Manager */ [data-state] { --_state---true: 1; --_state---false: 0; } .is-active, [data-state~="checked"]:is(:checked, :has(:checked)), [data-state~="current"]:is(.w--current, :has(.w--current)), [data-state~="open"]:is(.w--open, :has(.w--open)), [data-state~="expanded"]:is([aria-expanded="true"], :has([aria-expanded="true"])), [data-state~="external"]:is([target="_blank"], :has([target="_blank"])) { --_state---true: 0; --_state---false: 1; } .wf-design-mode [data-trigger~="preview"], [data-trigger~="focus"]:is(:focus-visible, :has(:focus-visible)), [data-trigger~="group"]:has([data-trigger~="focus-other"]:focus-visible, [data-trigger~="focus-other"] :focus-visible) [data-trigger~="focus-other"]:not(:focus-visible, :has(:focus-visible)) { --_trigger---on: 0; --_trigger---off: 1; } @media (hover: hover) { [data-button]:hover, [data-trigger~="hover"]:is(a:hover,button:hover,:has(a:hover,button:hover)), [data-trigger~="group"]:has([data-trigger~="hover-other"]:hover) [data-trigger~="hover-other"]:not(:hover) { --_trigger---on: 0; --_trigger---off: 1; } [data-trigger~="hover-other"]:hover { --_trigger---on: 1 !important; --_trigger---off: 0 !important; } } @media (hover: none) { [data-trigger~="mobile"] { --_trigger---on: 0; --_trigger---off: 1; } } code, kbd, pre, samp { font-family: var(--_typography---font--mono-family); } body * { scrollbar-width: none; /* Firefox */ -ms-overflow-style: none; /* IE/Edge Legacy */ } body *::-webkit-scrollbar { display: none; /* Unreliable */ width: 0px; /* WebKit/Blink */ } @media (prefers-color-scheme: dark) { body, .u-theme-ivory, [data-wf--section--theme="ivory"] { --_theme---background-primary: var(--swatch--gray-950); --_theme---background-secondary: var(--swatch--gray-900); --_theme---background-tertiary: var(--swatch--gray-850); --_theme---border-primary: var(--swatch--gray-600); --_theme---border-secondary: var(--swatch--gray-700); --_theme---border-tertiary: var(--swatch--gray-750); --_theme---foreground-primary: var(--swatch--gray-050); --_theme---foreground-secondary: var(--swatch--gray-400); --_theme---foreground-tertiary: var(--swatch--gray-500); --_theme---pictogram-accent: var(--swatch--gray-750); --_theme---button-primary--background: var(--swatch--gray-050); --_theme---button-primary--text: var(--swatch--gray-950); --_theme---button-primary--border: var(--swatch--transparent); --_theme---button-primary--icon: var(--_theme---button-primary--text); --_theme---button-primary--background-hover: var(--_theme---button-primary--background); --_theme---button-primary--text-hover: var(--_theme---button-primary--text); --_theme---button-primary--border-hover: var(--_theme---button-primary--border); --_theme---button-primary--icon-hover: var(--_theme---background-primary); --_theme---button-secondary--background: var(--swatch--gray-750); --_theme---button-secondary--text: var(--swatch--gray-050); --_theme---button-secondary--border: var(--_theme---border-secondary); --_theme---button-secondary--icon: var(--_theme---button-secondary--text); --_theme---button-secondary--background-hover: var(--_theme---button-secondary--background); --_theme---button-secondary--text-hover: var(--_theme---button-secondary--text); --_theme---button-secondary--border-hover: var(--_theme---button-secondary--background); --_theme---button-secondary--icon-hover: var(--_theme---foreground-secondary); --_theme---button-tertiary--background: var(--_theme---background-primary); --_theme---button-tertiary--text: var(--swatch--gray-050); --_theme---button-tertiary--border: var(--_theme---border-secondary); --_theme---button-tertiary--icon: var(--_theme---button-tertiary--text); --_theme---button-tertiary--background-hover: var(--_theme---button-tertiary--background); --_theme---button-tertiary--text-hover: var(--_theme---button-tertiary--text); --_theme---button-tertiary--border-hover: var(--_theme---button-tertiary--border); --_theme---button-tertiary--icon-hover: var(--_theme---foreground-primary); --_theme---error-text: #df6666; --_theme---heroes-accent: #C46849; --_theme---white: var(--_theme---background-primary); } .u-theme-white, [data-wf--section--theme="white"] { --_theme---background-primary: var(--swatch--gray-850); --_theme---background-secondary: var(--swatch--gray-800); --_theme---background-tertiary: var(--swatch--gray-750); --_theme---border-primary: var(--swatch--gray-550); --_theme---border-secondary: var(--swatch--gray-650); --_theme---border-tertiary: var(--swatch--gray-700); --_theme---foreground-primary: var(--swatch--gray-050); --_theme---foreground-secondary: var(--swatch--gray-350); --_theme---foreground-tertiary: var(--swatch--gray-450); --_theme---pictogram-accent: var(--swatch--gray-700); --_theme---button-primary--background: var(--swatch--gray-050); --_theme---button-primary--text: var(--swatch--gray-950); --_theme---button-primary--border: var(--swatch--transparent); --_theme---button-primary--icon: var(--_theme---button-primary--text); --_theme---button-primary--background-hover: var(--_theme---button-primary--background); --_theme---button-primary--text-hover: var(--_theme---button-primary--text); --_theme---button-primary--border-hover: var(--_theme---button-primary--border); --_theme---button-primary--icon-hover: var(--_theme---background-primary); --_theme---button-secondary--background: var(--swatch--gray-700); --_theme---button-secondary--text: var(--swatch--gray-050); --_theme---button-secondary--border: var(--_theme---border-secondary); --_theme---button-secondary--icon: var(--_theme---button-secondary--text); --_theme---button-secondary--background-hover: var(--_theme---button-secondary--background); --_theme---button-secondary--text-hover: var(--_theme---button-secondary--text); --_theme---button-secondary--border-hover: var(--_theme---button-secondary--background); --_theme---button-secondary--icon-hover: var(--_theme---foreground-secondary); --_theme---button-tertiary--background: var(--_theme---background-primary); --_theme---button-tertiary--text: var(--swatch--gray-050); --_theme---button-tertiary--border: var(--_theme---border-secondary); --_theme---button-tertiary--icon: var(--_theme---button-tertiary--text); --_theme---button-tertiary--background-hover: var(--_theme---button-tertiary--background); --_theme---button-tertiary--text-hover: var(--_theme---button-tertiary--text); --_theme---button-tertiary--border-hover: var(--_theme---button-tertiary--border); --_theme---button-tertiary--icon-hover: var(--_theme---foreground-primary); --_theme---error-text: #df6666; --_theme---heroes-accent: #C46849; --_theme---white: var(--_theme---background-primary); } .u-theme-neutral-1, [data-wf--section--theme="neutral-1"] { --_theme---background-primary: var(--swatch--gray-800); --_theme---background-secondary: var(--swatch--gray-750); --_theme---background-tertiary: var(--swatch--gray-700); --_theme---border-primary: var(--swatch--gray-500); --_theme---border-secondary: var(--swatch--gray-600); --_theme---border-tertiary: var(--swatch--gray-650); --_theme---foreground-primary: var(--swatch--gray-050); --_theme---foreground-secondary: var(--swatch--gray-300); --_theme---foreground-tertiary: var(--swatch--gray-400); --_theme---pictogram-accent: var(--swatch--gray-650); --_theme---button-primary--background: var(--swatch--gray-050); --_theme---button-primary--text: var(--swatch--gray-950); --_theme---button-primary--border: var(--swatch--transparent); --_theme---button-primary--icon: var(--_theme---button-primary--text); --_theme---button-primary--background-hover: var(--_theme---button-primary--background); --_theme---button-primary--text-hover: var(--_theme---button-primary--text); --_theme---button-primary--border-hover: var(--_theme---button-primary--border); --_theme---button-primary--icon-hover: var(--_theme---background-primary); --_theme---button-secondary--background: var(--swatch--gray-650); --_theme---button-secondary--text: var(--swatch--gray-050); --_theme---button-secondary--border: var(--_theme---border-secondary); --_theme---button-secondary--icon: var(--_theme---button-secondary--text); --_theme---button-secondary--background-hover: var(--_theme---button-secondary--background); --_theme---button-secondary--text-hover: var(--_theme---button-secondary--text); --_theme---button-secondary--border-hover: var(--_theme---button-secondary--background); --_theme---button-secondary--icon-hover: var(--_theme---foreground-secondary); --_theme---button-tertiary--background: var(--_theme---background-primary); --_theme---button-tertiary--text: var(--swatch--gray-050); --_theme---button-tertiary--border: var(--_theme---border-secondary); --_theme---button-tertiary--icon: var(--_theme---button-tertiary--text); --_theme---button-tertiary--background-hover: var(--_theme---button-tertiary--background); --_theme---button-tertiary--text-hover: var(--_theme---button-tertiary--text); --_theme---button-tertiary--border-hover: var(--_theme---button-tertiary--border); --_theme---button-tertiary--icon-hover: var(--_theme---foreground-primary); --_theme---error-text: #df6666; --_theme---heroes-accent: #C46849; --_theme---white: var(--_theme---background-primary); } .logo_light { display: none; } .logo_dark { display: block; } .illustration_light { display: none; } .illustration_dark { display: block; } } @media (prefers-color-scheme: light) { .logo_light { display: block; } .logo_dark { display: none; } .illustration_light { display: block; } .illustration_dark { display: none; } } .u-text-font-mono { --_text-style---trim-top: var(--_typography---font--mono-text-trim-top); --_text-style---trim-bottom: var(--_typography---font--mono-text-trim-bottom); } .u-checklist ul { list-style: none; margin: 0; padding: 0; } .u-checklist ul li { position: relative; padding-left: 2rem; } .u-checklist ul li::before { content: ""; position: absolute; left: 0; top: 0.1em; width: 1.5rem; height: 1.5rem; background-repeat: no-repeat; background-position: center; background-size: contain; background-image: url("data:image/svg+xml,%3Csvg%20width%3D%2224%22%20height%3D%2224%22%20viewBox%3D%220%200%2024%2024%22%20fill%3D%22none%22%20xmlns%3D%22http%3A//www.w3.org/2000/svg%22%3E%3Cpath%20d%3D%22M18.226%206.13068C18.4439%205.95655%2018.7615%205.95361%2018.9842%206.13888C19.2067%206.32458%2019.2604%206.63728%2019.1283%206.88304L19.0604%206.98382L10.0602%2017.784C9.95233%2017.9133%209.7949%2017.9908%209.62665%2017.9984C9.45844%2018.0059%209.29454%2017.9429%209.17547%2017.8238L4.97541%2013.6237L4.89806%2013.53C4.7446%2013.2971%204.7705%2012.9802%204.97541%2012.7753C5.18032%2012.5704%205.49726%2012.5445%205.73011%2012.698L5.82386%2012.7753L9.55868%2016.5101L18.1393%206.21506L18.226%206.13068Z%22%20fill%3D%22%235E5D59%22/%3E%3C/svg%3E"); } [class^="card_"][class$="_wrap"] .clickable_wrap.u-cover-absolute .clickable_link, [class^="card_"][class$="_wrap"] .clickable_wrap.u-cover-absolute .clickable_btn { outline-offset: var(--focus--offset-inner); } textarea[data-autogrow] { overflow-y: hidden; resize: none; height: 1.75rem; min-height: 0; } .btn_main_wrap::hover, .btn_small_wrap::hover, .btn_tiny_wrap::hover, .button_toggle_wrap::hover, .btn_icon_main_wrap::hover, .btn_icon_small_wrap::hover, .btn_icon_tiny_wrap::hover{ transition: /* Transition to click/active */ box-shadow ease-in-out 100ms, background ease-in-out 100ms, color ease-in-out 50ms; } .btn_main_wrap::active, .btn_small_wrap::active, .btn_tiny_wrap::active, .button_toggle_wrap::active, .btn_icon_main_wrap::active, .btn_icon_small_wrap::active, .btn_icon_tiny_wrap::active{ transition: /* Transition to click/active */ box-shadow ease-in-out 50ms, background ease-in-out 50ms, color ease-in-out 25ms; } .card_cs_grid_img img { max-width: 60%; max-height: 60%; } @container viewport (width < 30em) { [data-wf--grid--column-count="4"]:has(.card_feature_wrap) .c-grid { --_column-count---value: 1; } } @container viewport (min-width: 30em) and (max-width: 62em) { [data-wf--grid--column-count="4"]:has(.card_feature_wrap) .c-grid { --_column-count---value: 2; } } /* Mods for spacing and visibility of embed in accordian content used for schema */ .accordion_content_text p:has(+ .w-embed.w-script) { margin-bottom: 0; } /* Absolute inner SVG of lottie to prevent page jump */ .heroes_lottie_component svg { position: absolute; top: 0; left: 0; } /* Sticky scroll */ @media screen and (min-width: 992px) { .sticky_image_link_wrap:has(.sticky_image_link.w--current) { opacity: 1; width: calc((100% - var(--_grid---gutter)) * (6 / 12)); } } @media screen and (max-width: 767px) { .c-grid:last-child .sticky_image_block, .sticky_image_block:last-child { padding: 0; } .c-grid:last-child .sticky_image_wrap { margin-bottom: 0; } } #send, #threads, #get-help, #collaborate { display: block; /* or grid, flex - anything but contents */ } /* Select text below clickable overlay */ html.wf-design-mode .clickable_wrap { pointer-events: none; } document.addEventListener("DOMContentLoaded", function () { // ---------------- Config ---------------- const EDGE_PADDING = 16; // >= 1rem from edges const OFFSET_Y = 10; // gap under trigger const DIM_OPACITY = 0.3; const DIM_EASE_MS = 350; const CLOSE_DELAY = 120; const isCoarse = () => matchMedia("(hover: none), (pointer: coarse)").matches; // ---------------- Bubble (single instance) ---------------- function ensureBubble(){ let el = document.querySelector(".tt-bubble"); if (el) return el; el = document.createElement("div"); el.className = "tt-bubble u-theme-white"; el.setAttribute("role","tooltip"); el.setAttribute("aria-hidden","true"); el.style.left = "0px"; el.style.top = "0px"; el.innerHTML = `

`; document.body.appendChild(el); return el; } const bubble = ensureBubble(); const elH = bubble.querySelector("#tt-title"); const elB = bubble.querySelector("#tt-body"); const elClose = bubble.querySelector(".tt-close"); // ---------------- Parse [[term|heading|body]] anywhere ---------------- const TOKEN_RE = /\[\[([^|\]]+)\|([^|\]]+)\|([^\]]+)\]\]/g; const BLOCK_SKIP = new Set(["SCRIPT","STYLE","NOSCRIPT","TEXTAREA","INPUT","SELECT","CODE","PRE","TEMPLATE","IFRAME"]); function shouldSkipTextNode(n){ let el = n.parentElement; while (el){ if (BLOCK_SKIP.has(el.tagName) || el.isContentEditable) return true; el = el.parentElement; } return false; } const walker = document.createTreeWalker(document.body, NodeFilter.SHOW_TEXT); const textNodes = []; while (walker.nextNode()){ const n = walker.currentNode; if (!n.nodeValue || shouldSkipTextNode(n)) continue; if (TOKEN_RE.test(n.nodeValue)) textNodes.push(n); TOKEN_RE.lastIndex = 0; } textNodes.forEach(node => { const frag = document.createDocumentFragment(); const insideLink = !!node.parentElement.closest("a"); let text = node.nodeValue, last = 0; TOKEN_RE.lastIndex = 0; let m; while ((m = TOKEN_RE.exec(text))){ if (m.index > last) frag.appendChild(document.createTextNode(text.slice(last, m.index))); const term=m[1].trim(), heading=m[2].trim(), body=m[3].trim(); const t = insideLink ? document.createElement("span") : document.createElement("button"); if (insideLink){ t.setAttribute("role","button"); t.setAttribute("tabindex","0"); } else { t.type="button"; } t.className="tt-trigger"; t.textContent=term; t.setAttribute("data-tt-h", heading); t.setAttribute("data-tt-b", body); t.setAttribute("aria-haspopup","dialog"); t.setAttribute("aria-expanded","false"); frag.appendChild(t); last = TOKEN_RE.lastIndex; } if (last < text.length) frag.appendChild(document.createTextNode(text.slice(last))); node.parentNode.replaceChild(frag, node); }); // ---------------- State ---------------- let current = null; let hoverCount = 0; let closeTimer = null; // Dimming bookkeeping let dimCtx = null; // { container, dimEls:[], wrappedTexts:[], pathEls:[] } // ---------------- Find the correct "text element" container ---------------- function findTextContainer(trigger){ // Prefer common RTE wrappers let el = trigger.closest(".w-richtext, .rich-text, .rte, [data-rte]"); if (el) return el; // Otherwise climb until we find an ancestor that contains multiple block nodes anywhere inside. const BLOCK_SEL = "p,h1,h2,h3,h4,h5,h6,ul,ol,li,blockquote,pre,figure,figcaption"; el = trigger.parentElement; while (el && el !== document.body){ const blockCount = el.querySelectorAll(BLOCK_SEL).length; if (blockCount >= 2) return el; el = el.parentElement; } // Fallback: nearest non-inline container el = trigger.parentElement || document.body; while (el && el !== document.body){ const d = getComputedStyle(el).display; if (d !== "inline" && d !== "contents") return el; el = el.parentElement; } return document.body; } // Utility: child of `ancestor` that contains `target` (direct child) function directChildContaining(ancestor, target){ for (const ch of ancestor.children){ if (ch === target || ch.contains(target)) return ch; } return null; } function getElementTarget(e) { // If target is already an Element, use it if (e.target instanceof Element) return e.target; // Otherwise, walk the composed/path for the first Element const path = (typeof e.composedPath === 'function') ? e.composedPath() : []; for (const n of path) if (n instanceof Element) return n; return null; } // ---------------- Dim everything except the trigger branch (sibling branches only) ---------------- function dimAllOtherBranches(container, trigger){ undim(); // clear previous const dimEls = []; const wrappedTexts = []; const pathEls = []; // Build ELEMENT-only path [container -> ... -> trigger] const path = []; for (let el = trigger; el && el !== container; el = el.parentElement) path.push(el); path.push(container); path.reverse(); // At each ancestor level, find the *direct* child that leads to the trigger for (let i = 0; i < path.length; i++){ const anc = path[i]; const branchChild = (i < path.length - 1) ? directChildContaining(anc, path[i+1]) : path[i]; // last step is the trigger itself // Fade element siblings (whole branches) for (const child of anc.children){ if (child === branchChild) continue; // keep the path branch crisp // Never fade any element that is (or contains) the trigger if (child === trigger || child.contains(trigger)) continue; child.style.transition = `opacity ${DIM_EASE_MS}ms ease`; child.style.opacity = String(DIM_OPACITY); dimEls.push(child); } // Fade TEXT NODE siblings directly under this ancestor (outside branchChild) anc.childNodes.forEach(node => { if (node.nodeType !== 3) return; // text only if (!node.nodeValue || !node.nodeValue.trim()) return; // If this text node sits inside branchChild, skip if (branchChild && branchChild.contains && branchChild.contains(node)) return; const span = document.createElement("span"); span.style.transition = `opacity ${DIM_EASE_MS}ms ease`; span.style.opacity = String(DIM_OPACITY); span.textContent = node.nodeValue; node.parentNode.replaceChild(span, node); wrappedTexts.push(span); }); // Keep a reference to the path elements (so we can explicitly restore opacity if needed) if (anc && anc.nodeType === 1) pathEls.push(anc); } // Hard-guard: explicitly set opacity:1 on the entire path to neutralize any inherited fade pathEls.forEach(el => { el.style.opacity = "1"; }); dimCtx = { container, dimEls, wrappedTexts, pathEls }; } function undim(){ if (!dimCtx) return; const { dimEls, wrappedTexts, pathEls } = dimCtx; // Animate back dimEls.forEach(el => { el.style.transition = `opacity ${DIM_EASE_MS}ms ease`; el.style.opacity = "1"; // remove inline style after the animation so we don't override site CSS setTimeout(() => { if (el) el.style.opacity = ""; }, DIM_EASE_MS + 50); }); wrappedTexts.forEach(span => { span.style.transition = `opacity ${DIM_EASE_MS}ms ease`; span.style.opacity = "1"; span.addEventListener("transitionend", () => { if (!span.parentNode) return; span.parentNode.replaceChild(document.createTextNode(span.textContent || ""), span); }, { once:true }); }); // Clear hard-guard on path pathEls.forEach(el => { if (el) el.style.opacity = ""; }); dimCtx = null; } // ---------------- Positioning (centered, edge-aware, flip) ---------------- function clamp(v,min,max){ return Math.max(min,Math.min(max,v)); } function measureBubbleForPlacement(){ const wasOpen = bubble.classList.contains("is-open"); if (!wasOpen){ bubble.style.visibility="hidden"; bubble.classList.add("is-open"); } const rect = bubble.getBoundingClientRect(); if (!wasOpen){ bubble.classList.remove("is-open"); bubble.style.visibility=""; } return { w: rect.width, h: rect.height }; } function placeAnchored(trigger){ const vw=innerWidth, vh=innerHeight; const r = trigger.getBoundingClientRect(); const { w, h } = measureBubbleForPlacement(); let left = r.left + (r.width/2) - (w/2); left = clamp(left, EDGE_PADDING, Math.max(EDGE_PADDING, vw - EDGE_PADDING - w)); const topBelow = r.bottom + OFFSET_Y; const spaceBelow = vh - topBelow - EDGE_PADDING; const placeBelow = spaceBelow >= h; let top = placeBelow ? topBelow : (r.top - h - OFFSET_Y); top = clamp(top, EDGE_PADDING, Math.max(EDGE_PADDING, vh - EDGE_PADDING - h)); bubble.style.left = left + "px"; bubble.style.top = top + "px"; const br = bubble.getBoundingClientRect(); if (br.bottom > vh - EDGE_PADDING){ bubble.style.maxHeight = (vh - 2*EDGE_PADDING) + "px"; bubble.style.overflowY = "auto"; } else { bubble.style.maxHeight = "none"; bubble.style.overflowY = "visible"; } } // ---------------- Open / Close (place → fade/scale) ---------------- function animateIn(){ bubble.style.transition = "none"; bubble.style.opacity = "0"; bubble.style.transform = "scale(0.95)"; void bubble.offsetWidth; bubble.style.transition = "opacity .18s ease, transform .18s ease"; bubble.style.opacity = "1"; bubble.style.transform = "scale(1)"; } function animateOut(done){ bubble.style.transition = "opacity .16s ease, transform .16s ease"; bubble.style.opacity = "0"; bubble.style.transform = "scale(0.95)"; const end = () => { bubble.removeEventListener("transitionend", end); done && done(); }; bubble.addEventListener("transitionend", end); setTimeout(end, 260); } function openFromTrigger(trigger){ if (current && current !== trigger) forceClose(); current = trigger; trigger.setAttribute("aria-expanded","true"); elH.textContent = trigger.getAttribute("data-tt-h") || ""; elB.textContent = trigger.getAttribute("data-tt-b") || ""; bubble.classList.add("is-open"); bubble.setAttribute("aria-hidden","false"); placeAnchored(trigger); animateIn(); const container = findTextContainer(trigger); dimAllOtherBranches(container, trigger); hoverCount = 0; cancelCloseTimer(); } function forceClose(){ if (!current) return; bubble.classList.remove("is-open"); bubble.setAttribute("aria-hidden","true"); current.setAttribute("aria-expanded","false"); current = null; undim(); hoverCount = 0; cancelCloseTimer(); } function closeWithAnim(){ if (!current) return; const t = current; animateOut(() => { bubble.classList.remove("is-open"); bubble.setAttribute("aria-hidden","true"); t.setAttribute("aria-expanded","false"); current = null; undim(); }); } function scheduleClose(){ cancelCloseTimer(); closeTimer = setTimeout(() => { if (hoverCount <= 0 && !isCoarse()) closeWithAnim(); }, CLOSE_DELAY); } function cancelCloseTimer(){ if (closeTimer){ clearTimeout(closeTimer); closeTimer = null; } } // ---------------- Hover-intent (desktop) ---------------- function onZoneEnter(){ if (isCoarse()) return; hoverCount++; cancelCloseTimer(); } function onZoneLeave(){ if (isCoarse()) return; hoverCount = Math.max(0, hoverCount - 1); if (hoverCount === 0) scheduleClose(); } bubble.addEventListener("pointerenter", onZoneEnter, true); bubble.addEventListener("mouseenter", onZoneEnter, true); bubble.addEventListener("pointerleave", onZoneLeave, true); bubble.addEventListener("mouseleave", onZoneLeave, true); const handleEnter = (e) => { if (isCoarse()) return; const target = getElementTarget(e); if (!target) return; const t = target.closest(".tt-trigger"); if (!t) return; onZoneEnter(); if (!current || current !== t) openFromTrigger(t); }; const handleLeave = (e) => { if (isCoarse()) return; const target = getElementTarget(e); if (!target) return; const t = target.closest(".tt-trigger"); if (!t) return; onZoneLeave(); }; document.addEventListener("pointerenter", handleEnter, true); document.addEventListener("mouseenter", handleEnter, true); document.addEventListener("pointerleave", handleLeave, true); document.addEventListener("mouseleave", handleLeave, true); // ---------------- Keyboard ---------------- document.addEventListener("focusin", (e) => { if (!e.target) return; const t = e.target.closest(".tt-trigger"); if (t) openFromTrigger(t); }); document.addEventListener("focusout", (e) => { if (!e.target) return; const t = e.target.closest(".tt-trigger"); if (t && current === t) closeWithAnim(); }); // ---------------- Mobile / coarse ---------------- document.addEventListener("pointerdown", (e) => { if (!isCoarse()) return; const t = e.target.closest(".tt-trigger"); if (!t) return; e.preventDefault(); e.stopPropagation(); if (current === t && bubble.classList.contains("is-open")) { closeWithAnim(); return; } openFromTrigger(t); }, true); document.addEventListener("click", (e) => { if (!isCoarse()) return; if (!bubble.classList.contains("is-open")) return; const inBubble = !!e.target.closest(".tt-bubble"); const onTrigger = !!e.target.closest(".tt-trigger"); if (!inBubble && !onTrigger) closeWithAnim(); }, true); // Close button + ESC elClose.addEventListener("click", closeWithAnim); document.addEventListener("keydown", (e) => { if (e.key === "Escape") closeWithAnim(); }); // Reposition on resize/scroll while open const reposition = () => { if (!current) return; placeAnchored(current); }; addEventListener("resize", reposition, { passive: true }); addEventListener("scroll", reposition, { passive: true }); }); /* Tooltip Styles */ /* Trigger */ .tt-trigger { cursor: help; text-decoration: underline dotted; text-underline-offset: .2em; color: inherit; } .tt-trigger:focus-visible{ outline:2px solid currentColor; outline-offset:2px; } /* Bubble */ .tt-bubble{ position: fixed; z-index: 10; max-width: 17rem; background: var(--_theme---background-primary); box-shadow: 0 4px 24px rgba(0,0,0,.05); border-radius: var(--radius--large); border-style: solid; border-color: var(--_theme---border-tertiary); padding: var(--_spacing---space--1-5rem); pointer-events: none; opacity: 0; transform: translate3d(0,0,0) scale(.98); transition: opacity .3s ease, transform .3s ease; will-change: transform, opacity; } .tt-bubble.is-open{ opacity:1; transform:translate3d(0,0,0) scale(1); pointer-events:auto; } .tt-h{ margin-bottom: var(--_spacing---space--0-5rem); font-size: var(--_typography---font-size--body-3); font-family: var(--_typography---font--secondary-family); line-height: var(--_typography---line-height--1-6); color: var(--_theme---foreground-primary); } .tt-b{ margin:0; font-size: var(--_typography---font-size--caption); line-height: var(--_typography---line-height--1-6); color: var(--_theme---foreground-tertiary); } /* Mobile close button */ .tt-close { display: none; } @media (hover: none), (pointer: coarse) { .tt-close { display: inline-flex; position: absolute; top: 0.5rem; right: 0.5rem; width: 32px; height: 32px; align-items: center; justify-content: center; border: 0; border-radius: 999px; background: transparent; font-size: 22px; line-height: 1; color: inherit; cursor: pointer; touch-action: manipulation; } .tt-close:focus-visible { outline: 2px solid currentColor; outline-offset: 2px; } .tt-close:hover { opacity: 1; } } html[lang="de-DE"] h1, html[lang="de-DE"] h2, html[lang="de-DE"] h3, html[lang="de-DE"] h4, html[lang="de-DE"] h5, html[lang="de-DE"] h6, html[lang="de-DE"] p, html[lang="de-DE"] li, html[lang="fr-FR"] h1, html[lang="fr-FR"] h2, html[lang="fr-FR"] h3, html[lang="fr-FR"] h4, html[lang="fr-FR"] h5, html[lang="fr-FR"] h6, html[lang="fr-FR"] p, html[lang="fr-FR"] li { overflow-wrap: break-word; hyphens: auto; } :root { --nav--icon-thickness: var(--border-width--main); --nav--hamburger-thickness: var(--nav--icon-thickness); --nav--hamburger-gap: var(--_spacing---space--0-25rem); --nav--hamburger-rotate: 45; --nav--dropdown-duration: 300ms; --nav--dropdown-open-duration: 600ms; --nav--dropdown-delay: 0ms; --ease-expo-out: cubic-bezier(0.16, 1, 0.3, 1); } /* ========== GENERAL RESPONSIVE RULES ========== */ /* Lock body when nav is open (script toggles .is-nav-open) */ @media (width < 56em) { body.is-nav-open { overflow: hidden; } } @container (min-width: 56em) { .nav_wrap.is-desktop { display: block; } .nav_wrap.is-mobile { display: none; } } @container (width < 28em) { .nav_mobile_layout .nav_actions_mobile { display: none; } } /* ========== DROPDOWN STYLING ========== */ html:not(.wf-design-mode) .nav_dropdown_component > .w-dropdown-list { display: grid !important; grid-template-columns: minmax(0, 1fr); grid-template-rows: 0fr; transition: grid-template-rows var(--nav--dropdown-duration) var(--ease-expo-out), visibility 0s var(--nav--dropdown-duration), opacity var(--nav--dropdown-duration) var(--ease-expo-out); visibility: hidden; opacity: 0; } html:not(.wf-design-mode) .nav_dropdown_component > .w-dropdown-list.w--open { visibility: visible; opacity: 1; transition: grid-template-rows var(--nav--dropdown-duration) var(--ease-expo-out), visibility 0s 0s, opacity var(--nav--dropdown-duration) var(--ease-expo-out); } .nav_dropdown_component > .w-dropdown-list > * { overflow: hidden; } .nav_dropdown_component:has(> .w-dropdown-toggle[aria-expanded="true"]) > .w-dropdown-list { --nav--dropdown-duration: var(--nav--dropdown-open-duration); grid-template-rows: 1fr; } /*.nav_wrap.is-desktop:has(.nav_dropdown_component > .w-dropdown-toggle.w--open[aria-expanded="false"]) .nav_dropdown_component:has(> .w--open[aria-expanded="true"]) > .w-dropdown-list { transition-delay: var(--nav--dropdown-duration); }*/ /* Dropdown caret rotation */ .nav_links_svg.is-desktop { transition: transform 750ms var(--ease-expo-out); } .w-dropdown-toggle[aria-expanded="true"] .nav_links_svg.is-desktop { transform: rotate(-180deg); } .nav_links_svg_line.is-2 { transition: transform 500ms var(--ease-expo-out); } .w-dropdown-toggle[aria-expanded="true"] .nav_links_svg_line.is-2 { transform: rotate(0deg); } /* open (replicates your original transforms) */ .nav_btn_wrap[aria-expanded="true"] .nav_btn_line:nth-child(1), .nav_btn_wrap[aria-expanded="true"] > * > :first-child { transform: translateY(calc(var(--nav--hamburger-thickness) * 1 + var(--nav--hamburger-gap) * 1)) rotate(calc(var(--nav--hamburger-rotate) * -1deg)); } .nav_btn_wrap[aria-expanded="true"] .nav_btn_line:nth-child(2), .nav_btn_wrap[aria-expanded="true"] > * > :nth-child(2) { opacity: 0; } .nav_btn_wrap[aria-expanded="true"] .nav_btn_line:nth-child(3), .nav_btn_wrap[aria-expanded="true"] > * > :last-child { transform: translateY(calc(var(--nav--hamburger-thickness) * -1 + var(--nav--hamburger-gap) * -1)) rotate(calc(var(--nav--hamburger-rotate) * 1deg)); width: 1rem; } /* ========== HOVER & THEME EFFECTS ========== */ @media (hover: hover) and (pointer: fine) { body:has(.nav_dropdown_item:hover) .nav_dropdown_item:not(:hover) > * > * { color: var(--_theme---foreground-tertiary); } .nav_dropdown_link { transition: background-color 300ms ease, color 300ms ease; } .nav_dropdown_item:hover .nav_dropdown_link { background: var(--_theme---background-tertiary); color: var(--_theme---foreground-primary); } .nav_secondary_wrap .nav_dropdown_item:hover .nav_dropdown_link { background: var(--_theme---background-tertiary); color: var(--_theme---foreground-primary); } .nav_wrap.is-mobile .nav_dropdown_item:hover .nav_dropdown_link { color: var(--_theme---foreground-primary); } .nav_links_text { transition: color 500ms var(--ease-expo-out); } .nav_links_svg { transition: transform 500ms var(--ease-expo-out), color 500ms var(--ease-expo-out); } .nav_links_item:hover .nav_links_text, .nav_links_item:hover .nav_links_svg { color: var(--_theme---foreground-primary); } } /* ========== LAYOUT / UTILITY (kept, de-Webflow’d) ========== */ .nav_wrap.is-mobile [data-open-on-mobile] > .w-dropdown-toggle { display: none; } .nav_wrap.is-mobile [data-open-on-mobile] > .w-dropdown-list { visibility: visible; opacity: 1; display: block; grid-template-rows: 1fr; } .nav_buttons_item .button_main_wrap { width: 100%; min-width: max-content; } /* Optional: fade out mobile actions while open */ .nav_actions_wrap { transition: opacity 500ms var(--ease-expo-out); } body.is-nav-open .nav_actions_mobile .nav_actions_wrap { opacity: 0; pointer-events: none; } .nav_links_item:first-child { border-top: none; } /* ========== Breadcrumbs ========== */ .breadcrumb_text.is_linked[href="#"], .breadcrumb_text:has(+.breadcrumb_text.is_linked:not([href="#"])) { display: none; } .breadcrumb_text.is_linked:not([href="#"]) { display: block; } document.addEventListener("DOMContentLoaded", function () { document.querySelectorAll(".nav_component").forEach((root) => { if (root.dataset.scriptInitialized) return; root.dataset.scriptInitialized = "true"; if (!window.gsap) { console.error("GSAP not found"); return; } const btn = root.querySelector('.nav_btn_wrap'); const menu = root.querySelector('.nav_menu_wrap'); if (!btn || !menu) { console.warn('Missing .nav_btn_wrap or .nav_menu_wrap in', root); return; } // a11y setup (scoped) if (!btn.hasAttribute('type')) btn.setAttribute('type', 'button'); if (!btn.hasAttribute('aria-expanded')) btn.setAttribute('aria-expanded', 'false'); if (!menu.id) menu.id = 'primary-nav-' + Math.random().toString(36).slice(2); if (!btn.hasAttribute('aria-controls')) btn.setAttribute('aria-controls', menu.id); menu.setAttribute('aria-hidden', 'true'); // targets (scoped) const items = Array.from(menu.querySelectorAll('.nav_links_item')); const actions = menu.querySelector('.nav_menu_actions_wrap'); // feature detection const prefersReduced = window.matchMedia('(prefers-reduced-motion: reduce)').matches; const canClipInset = !!(window.CSS && CSS.supports && ( CSS.supports('clip-path','inset(0 0 100% 0)') || CSS.supports('-webkit-clip-path','inset(0 0 100% 0)') )); const useClip = canClipInset && !prefersReduced; // durations from CSS vars function readDur(varName, fallbackSec) { const v = getComputedStyle(document.documentElement).getPropertyValue(varName).trim(); if (!v) return fallbackSec; if (v.endsWith('ms')) return parseFloat(v)/1000; if (v.endsWith('s')) return parseFloat(v); const n = parseFloat(v); return isNaN(n) ? fallbackSec : n; } const OPEN_DUR = readDur('--nav--menu-open-duration', 0.8); const CLOSE_DUR = readDur('--nav--menu-close-duration', 0.4); // state per component let isOpen = false; let current = null; function setMenuVisibleForAnim() { menu.style.display = 'flex'; // ensure it's shown before anim menu.removeAttribute('hidden'); menu.setAttribute('aria-hidden', 'false'); menu.style.willChange = useClip ? 'clip-path' : 'transform, opacity'; } function clearMenuInline() { gsap.set(menu, { clearProps: 'clipPath,webkitClipPath,opacity,transform,willChange,pointerEvents' }); } // OPEN function playOpen() { setMenuVisibleForAnim(); menu.style.pointerEvents = 'none'; document.body.classList.add('is-nav-open'); btn.setAttribute('aria-expanded', 'true'); if (useClip) { gsap.set(menu, { clipPath: 'inset(0 0 100% 0)', webkitClipPath: 'inset(0 0 100% 0)' }); } else { gsap.set(menu, { yPercent: -2, opacity: 0 }); } if (items.length) gsap.set(items, { y: 20, autoAlpha: 0 }); if (actions) gsap.set(actions, { y: 20, autoAlpha: 0 }); const tl = gsap.timeline({ defaults: { ease: 'power3.out' } }); if (useClip) { tl.to(menu, { clipPath: 'inset(0 0 0% 0)', webkitClipPath: 'inset(0 0 0% 0)', duration: prefersReduced ? 0.01 : OPEN_DUR, ease: 'expo.out' }, 0); } else { tl.to(menu, { yPercent: 0, opacity: 1, duration: prefersReduced ? 0.01 : Math.min(OPEN_DUR, 0.36) }, 0.02); } if (items.length) { tl.to(items, { y: 0, autoAlpha: 1, stagger: prefersReduced ? 0 : 0.08, duration: prefersReduced ? 0.01 : 0.4 }, 0.10); } if (actions) { const base = 0.10 + (items.length ? items.length * (prefersReduced ? 0 : 0.08) : 0); tl.to(actions, { y: 0, autoAlpha: 1, duration: prefersReduced ? 0.01 : 0.4 }, base); } tl.add(() => { menu.style.pointerEvents = 'auto'; }, '>-0.1'); // listeners (per open) document.addEventListener('keydown', onKeydown); menu.addEventListener('click', onMenuLinkClick); return tl; } // CLOSE (fade all together, then clip inset close) function playClose() { menu.style.pointerEvents = 'none'; btn.setAttribute('aria-expanded', 'false'); document.body.classList.remove('is-nav-open'); const tl = gsap.timeline({ defaults: { ease: 'power3.out' } }); const fadeTargets = items.concat(actions ? [actions] : []); if (fadeTargets.length) tl.to(fadeTargets, { autoAlpha: 0, y: 0, duration: prefersReduced ? 0.01 : 0.2 }, 0); if (useClip) { gsap.set(menu, { clipPath: 'inset(0 0 0% 0)', webkitClipPath: 'inset(0 0 0% 0)' }); tl.to(menu, { clipPath: 'inset(0 0 100% 0)', webkitClipPath: 'inset(0 0 100% 0)', duration: prefersReduced ? 0.01 : CLOSE_DUR }, '>-0.02'); } else { tl.to(menu, { yPercent: -2, opacity: 0, duration: prefersReduced ? 0.01 : Math.min(CLOSE_DUR, 0.28) }, '>-0.02'); } tl.add(() => { menu.style.display = 'none'; menu.setAttribute('aria-hidden', 'true'); clearMenuInline(); if (items.length) gsap.set(items, { clearProps: 'all' }); if (actions) gsap.set(actions, { clearProps: 'all' }); // remove listeners added on open document.removeEventListener('keydown', onKeydown); menu.removeEventListener('click', onMenuLinkClick); }); return tl; } function openMenu() { if (isOpen) return; isOpen = true; if (current && current.isActive()) current.kill(); current = playOpen(); } function closeMenu() { if (!isOpen) return; isOpen = false; if (current && current.isActive()) current.kill(); current = playClose(); } function onKeydown(e){ if (e.key === 'Escape' && isOpen) { e.preventDefault(); closeMenu(); } } function onMenuLinkClick(e){ const a = e.target.closest('a[href]'); if (!a) return; const url = new URL(a.href, location.href); if (url.origin === location.origin) { e.preventDefault(); const tl = playClose(); tl.eventCallback('onComplete', () => { window.location.href = a.href; }); isOpen = false; } } // Toggle (scoped) btn.addEventListener('click', () => (isOpen ? closeMenu() : openMenu())); // Normalize if visible on load (scoped) if (getComputedStyle(menu).display !== 'none') { btn.setAttribute('aria-expanded', 'true'); menu.setAttribute('aria-hidden', 'false'); document.body.classList.add('is-nav-open'); isOpen = true; } }); }); (function () { 'use strict'; // ---------- tiny utils ---------- var NS = 'navBundleInit'; function onceFlag(el, k){ k=k||'scriptInitialized'; if (el.dataset[k]) return true; el.dataset[k]='true'; return false; } function ready(fn){ if(document.readyState==='loading'){document.addEventListener('DOMContentLoaded',fn,{once:true});} else { fn(); } } // ---------- 1) Ask Claude about this page ---------- function initAskPage(){ var buttons = document.querySelectorAll('[data-ask-page]'); if (!buttons.length) return; // Early exit if no buttons buttons.forEach(function(btn){ if (onceFlag(btn, NS)) return; btn.addEventListener('click', function (e) { e.preventDefault(); var pageUrl = window.location.href; var prompt = "Read this page " + pageUrl + " so that I can ask you questions about it"; var claudeUrl = new URL('https://claude.ai/new'); claudeUrl.searchParams.set('q', prompt); window.open(claudeUrl.toString(), '_blank', 'noopener'); }); }); } // ---------- 2) Copy page content as Markdown (Turndown) ---------- var _turndownReady; function ensureTurndown(){ if (window.TurndownService) return Promise.resolve(); if (_turndownReady) return _turndownReady; _turndownReady = new Promise(function(resolve, reject){ var s = document.createElement('script'); s.src = 'https://unpkg.com/turndown/dist/turndown.js'; s.async = true; s.onload = function(){ resolve(); }; s.onerror = function(){ reject(new Error('Failed to load Turndown')); }; document.head.appendChild(s); }); return _turndownReady; } function initCopyAsMarkdown(){ var copyButton = document.getElementById('copy-as-markdown'); if (!copyButton) return; // Early exit if (onceFlag(copyButton, NS)) return; var buttonTextEl = copyButton.querySelector('.nav_dropdown_text') || copyButton; var originalText = buttonTextEl.textContent; copyButton.addEventListener('click', function(){ ensureTurndown().then(function(){ try { var TurndownService = window.TurndownService; var turndownService = new TurndownService({ headingStyle: 'atx', codeBlockStyle: 'fenced', fence: '```', emDelimiter: '*', strongDelimiter: '**', linkStyle: 'inlined' }); // Skip junk turndownService.addRule('skipWebflowElements', { filter: function(node){ return node.nodeName === 'SCRIPT' || node.nodeName === 'STYLE' || (node.className && (String(node.className).includes('w-editor') || String(node.className).includes('w-embed'))); }, replacement: function(){ return ''; } }); buttonTextEl.textContent = 'Copying...'; copyButton.disabled = true; var contentElement = document.querySelector('main') || document.querySelector('.main-content') || document.querySelector('body'); if (!contentElement) throw new Error('No content found to copy'); var cloned = contentElement.cloneNode(true); cloned.querySelectorAll('script, style, nav, footer, .w-nav, .footer').forEach(function(el){ el.remove(); }); var markdown = turndownService.turndown(cloned); var done = function(){ buttonTextEl.textContent = 'Copied!'; setTimeout(function(){ buttonTextEl.textContent = originalText; copyButton.disabled = false; }, 2000); }; if (navigator.clipboard && window.isSecureContext) { navigator.clipboard.writeText(markdown).then(done, function(err){ throw err; }); } else { var ta = document.createElement('textarea'); ta.value = markdown; ta.style.position='fixed'; ta.style.opacity='0'; document.body.appendChild(ta); ta.select(); document.execCommand('copy'); document.body.removeChild(ta); done(); } } catch (err){ console.error('Copy failed:', err); buttonTextEl.textContent = 'Copy failed'; setTimeout(function(){ buttonTextEl.textContent = originalText; copyButton.disabled = false; }, 2000); } }).catch(function(err){ console.error('Turndown load failed:', err); buttonTextEl.textContent = 'Copy failed'; setTimeout(function(){ buttonTextEl.textContent = originalText; copyButton.disabled = false; }, 2000); }); }); } // ---------- init all ---------- ready(function(){ initAskPage(); initCopyAsMarkdown(); }); // Optional: expose minimal API for debugging (comment out in production if not needed) // window.NavBundle = { // initCopyAsMarkdown: initCopyAsMarkdown, // initAskPage: initAskPage // }; })(); Explore here /* - - - - - - - - - - - - - - - - - - - - - - - - - - - THUMBNAIL BG COLORS - - - - - - - - - - - - - - - - - - - - - - - - - - - - */ [data-thumbnail-bg="Clay"] { background-color: var(--swatch--clay); } [data-thumbnail-bg="Cactus"] { background-color: var(--swatch--cactus); } [data-thumbnail-bg="Coral"] { background-color: var(--swatch--coral); } [data-thumbnail-bg="Fig"] { background-color: var(--swatch--fig); } [data-thumbnail-bg="Heather"] { background-color: var(--swatch--heather); } [data-thumbnail-bg="Oat"] { background-color: var(--swatch--oat); } [data-thumbnail-bg="Olive"] { background-color: var(--swatch--olive); } [data-thumbnail-bg="Sky"] { background-color: var(--swatch--sky); } /* - - - - - - - - - - - - - - - - - - - - - - - - - - - CONDITIONAL STYLES BASED ON IF IMAGE OR VIDEO ARE VISIBLE - - - - - - - - - - - - - - - - - - - - - - - - - - - - */ .hero_tutorial_post_wrap:has(+ .tutorial_post_media_wrap :is(.tutorial_post_image_wrap, .tutorial_post_video_wrap):not(.w-condition-invisible)) { border-bottom: none !important; } .tutorial_post_media_wrap:has(+ .tutorial_post_wrap:not(.w-condition-invisible)) .u-section-spacer { display: none !important; } /* - - - - - - - - - - - - - - - - - - - - - - - - - - - TABLE OF CONTENTS Adding here so classes don't get accidentally cleared - - - - - - - - - - - - - - - - - - - - - - - - - - - - */ .tutorial_post_toc_list { color: var(--_theme---foreground-tertiary); font-size: var(--_typography---font-size--body-3); } .tutorial_post_toc_item { margin-left: -0.0625rem; padding-top: var(--_spacing---space--0-25rem); padding-bottom: var(--_spacing---space--0-25rem); padding-left: var(--_spacing---space--1rem); border-left-style: solid; border-left-width: var(--border-width--main); border-left-color: var(--_theme---border-secondary); } .tutorial_post_toc_item.active { border-left-color: var(--_theme---heroes-accent); } /* - - - - - - - - - - - - - - - - - - - - - - - - - - - ANCHOR LINK - - - - - - - - - - - - - - - - - - - - - - - - - - - - */ /* offset for that takes fixed header height into account */ .anchor { position: relative; top: calc(-1 * var(--_spacing---space--4rem) - 120px); } /* Copy anchor link to clipboard button */ .anchor-link-box { /* quick btn reset */ background: none; border: none; padding: 0; cursor: pointer; /* other styles */ position: absolute !important; left: -3rem; top: 50%; transform: translateY(-50%); opacity: 0; transition: opacity 0.2s ease; width: 2.5rem; height: 2.5rem; background: var(--_theme---background-tertiary); border: var(--border-width--main) solid var(--_theme---border-tertiary); border-radius: var(--radius--main); display: flex; align-items: center; justify-content: center; padding: var(--_spacing---space--0-5rem); } .anchor-link-box:hover { opacity: 1; } .tooltip { opacity: 0; transition: opacity 0.2s ease; position: absolute; left: 50%; bottom: calc(100% + 0.5rem); transform: translateX(-50%); background: var(--_theme---button-primary--background); color: var(--_theme---button-primary--text); white-space: nowrap; padding: 0.125rem 0.5rem; border-radius: var(--radius--x-small); font-family: var(--_typography---font--primary-family); font-size: var(--_typography---font-size--caption); line-height: var(--_typography---line-height--1-6); } .tooltip.active { opacity: 1; } /* - - - - - - - - - - - - - - - - - - - - - - - - - - - HIDE EMPTY

TAGS - - - - - - - - - - - - - - - - - - - - - - - - - - - - */ p:empty, p:has(> br:only-child) { display: none; } /* - - - - - - - - - - - - - - - - - - - - - - - - - - - BORDER RADIUS ON RICH TEXT VIDEOS - - - - - - - - - - - - - - - - - - - - - - - - - - - - */ .w-richtext-figure-type-video { border-radius: var(--radius--main); overflow: hidden; } /* - - - - - - - - - - - - - - - - - - - - - - - - - - - FIX GRID SPACING IF TUTORIAL HAS NO VIDEO - - - - - - - - - - - - - - - - - - - - - - - - - - - - */ .tutorial_post_video_layout:has(.w-dyn-bind-empty) { display: none; } document.addEventListener('DOMContentLoaded', function() { // Helper function to convert text to slug function textToSlug(text) { return text .toLowerCase() .trim() .replace(/[''"&]/g, '') .replace(/"([^"]+)"/, '$1') .replace(/\s*[:.,?—–\\/-]\s*/g, '-') .replace(/\s+/g, '-') .replace(/-+/g, '-') .replace(/^-+|-+$/g, ''); } // Track used slugs to prevent duplicate IDs const usedSlugs = new Set(); function getUniqueSlug(baseSlug) { let slug = baseSlug; let counter = 1; while (usedSlugs.has(slug)) { slug = `${baseSlug}-${counter}`; counter++; } usedSlugs.add(slug); return slug; } // Add/Remove class helper functions const addClass = (element, className) => { if (element && className) { element.classList.add(className); } }; const removeClass = (element, className) => { if (element && className) { element.classList.remove(className); } }; // Debounce helper for resize events function debounce(func, wait) { let timeout; return function executedFunction(...args) { const later = () => { clearTimeout(timeout); func(...args); }; clearTimeout(timeout); timeout = setTimeout(later, wait); }; } // Get elements const tutorialContent = document.getElementById('tutorial_content'); const tocContainer = document.getElementById('tutorial_toc'); if (!tutorialContent || !tocContainer) { return; } const h2Elements = tutorialContent.querySelectorAll('h2'); if (h2Elements.length === 0) { tocContainer.style.display = 'none'; return; } // Create ToC list const tocList = document.createElement('ul'); addClass(tocList, 'tutorial_post_toc_list'); const tocItems = []; // Copy anchor link to clipboard with tooltip const copyAnchorToClipboard = (element, spanElement, mouseEnterTime = 750) => { let hoverTimeout; let tooltip; element.addEventListener("mouseenter", function () { clearTimeout(hoverTimeout); tooltip = document.createElement("span"); addClass(tooltip, 'tooltip'); tooltip.innerText = "Copy anchor link"; this.appendChild(tooltip); hoverTimeout = setTimeout(() => { addClass(tooltip, "active"); }, mouseEnterTime); }); element.addEventListener("mouseleave", function () { clearTimeout(hoverTimeout); if (tooltip) { removeClass(tooltip, "active"); this.removeChild(tooltip); tooltip = null; } }); element.addEventListener("click", function (event) { event.preventDefault(); clearTimeout(hoverTimeout); if (spanElement && spanElement.classList.contains("anchor")) { const id = spanElement.id; if (tooltip) { tooltip.innerText = "Copied!"; addClass(tooltip, "active"); const url = new URL(window.location.href); url.searchParams.delete("topic"); const anchorLink = `${url.origin}${url.pathname}#${id}`; if (navigator && navigator.clipboard) { navigator.clipboard.writeText(anchorLink).catch((err) => { console.error("Could not copy text: ", err); }); } hoverTimeout = setTimeout(() => { removeClass(tooltip, "active"); }, 1500); } } else { console.error("Couldn't find Span with the ID"); } }); }; // Anchor link SVG icon const anchorLinkSVG = ``; // Process each H2 and build ToC h2Elements.forEach((h2) => { const headingText = h2.textContent; const baseSlug = textToSlug(headingText); const slug = getUniqueSlug(baseSlug); // Add anchor span const anchorSpan = document.createElement('span'); anchorSpan.id = slug; addClass(anchorSpan, 'anchor'); h2.prepend(anchorSpan); // Add anchor link box const anchorLinkBox = document.createElement('button'); anchorLinkBox.type = 'button'; addClass(anchorLinkBox, 'anchor-link-box'); anchorLinkBox.innerHTML = anchorLinkSVG; h2.prepend(anchorLinkBox); // Initialize copy functionality copyAnchorToClipboard(anchorLinkBox, anchorSpan); // Create ToC item const listItem = document.createElement('li'); addClass(listItem, 'tutorial_post_toc_item'); const link = document.createElement('a'); link.href = `#${slug}`; const textDiv = document.createElement('div'); textDiv.textContent = headingText; link.appendChild(textDiv); listItem.appendChild(link); tocList.appendChild(listItem); tocItems.push({ element: listItem, target: h2 }); }); tocContainer.appendChild(tocList); // Configuration const headerHeight = 134; const activationOffset = 100; // State let currentActive = 0; let ranges = []; let frameRequested = false; // Calculate activation ranges using getBoundingClientRect for reliability // (offsetTop can be relative to offsetParent, not document) function calculateRanges() { const scrollY = window.scrollY; ranges = tocItems.map((item, index) => { const rect = item.target.getBoundingClientRect(); const absoluteTop = rect.top + scrollY; const activationPoint = absoluteTop - headerHeight - activationOffset; const nextAbsoluteTop = index < tocItems.length - 1 ? tocItems[index + 1].target.getBoundingClientRect().top + scrollY : Infinity; const nextActivation = nextAbsoluteTop === Infinity ? Infinity : nextAbsoluteTop - headerHeight - activationOffset; return { index: index, start: Math.max(0, activationPoint), end: nextActivation }; }); } // Update active section function updateActiveSection() { if (ranges.length === 0) return; const scrollPosition = window.scrollY; let newActive = 0; // Find current range if (scrollPosition < ranges[0].start) { newActive = 0; } else { for (let i = 0; i < ranges.length; i++) { if (scrollPosition >= ranges[i].start && scrollPosition < ranges[i].end) { newActive = i; break; } } } // Special case: if near bottom of page, activate last item const scrollBottom = scrollPosition + window.innerHeight; const documentHeight = document.documentElement.scrollHeight; if (documentHeight - scrollBottom < 100) { newActive = tocItems.length - 1; } // Update classes if changed if (newActive !== currentActive) { tocItems.forEach(item => removeClass(item.element, 'active')); currentActive = newActive; if (tocItems[currentActive]) { addClass(tocItems[currentActive].element, 'active'); } } frameRequested = false; } // Scroll handler function handleScroll() { if (!frameRequested) { frameRequested = true; window.requestAnimationFrame(updateActiveSection); } } // Recalculate on resize (debounced) const handleResize = debounce(() => { calculateRanges(); updateActiveSection(); }, 250); // Smooth scroll behavior for ToC links tocItems.forEach(item => { item.element.querySelector('a').addEventListener('click', function(e) { e.preventDefault(); const targetId = this.getAttribute('href').substring(1); const targetElement = document.getElementById(targetId); if (targetElement) { const scrollTo = targetElement.parentElement.offsetTop - headerHeight - 20; window.scrollTo({ top: scrollTo, behavior: 'smooth' }); } }); }); // Initialize calculateRanges(); addClass(tocItems[0].element, 'active'); // Handle hash on initial page load if (window.location.hash) { const hash = window.location.hash.substring(1); const targetIndex = tocItems.findIndex(item => item.target.querySelector('.anchor')?.id === hash ); if (targetIndex !== -1) { tocItems.forEach(item => removeClass(item.element, 'active')); addClass(tocItems[targetIndex].element, 'active'); currentActive = targetIndex; } } // Re-check after browser might restore scroll position requestAnimationFrame(() => { requestAnimationFrame(() => { calculateRanges(); updateActiveSection(); }); }); // Event listeners window.addEventListener('scroll', handleScroll, { passive: true }); window.addEventListener('resize', handleResize); // Visual Viewport API for mobile viewport changes (keyboard, browser chrome) if (window.visualViewport) { window.visualViewport.addEventListener('resize', handleResize); window.visualViewport.addEventListener('scroll', handleResize); } // Browser back/forward navigation with hash changes window.addEventListener('popstate', () => { calculateRanges(); updateActiveSection(); }); // Handle dynamic content changes (if Webflow adds/removes content) const observer = new MutationObserver(debounce(() => { calculateRanges(); updateActiveSection(); }, 500)); observer.observe(tutorialContent, { childList: true, subtree: true, attributes: false }); // Recalculate after images load (affects layout positions) const images = tutorialContent.querySelectorAll('img'); images.forEach(img => { if (!img.complete) { img.addEventListener('load', handleResize, { once: true }); img.addEventListener('error', handleResize, { once: true }); } }); // Recalculate after fonts load if (document.fonts && document.fonts.ready) { document.fonts.ready.then(handleResize); } }); How to use Claude in Excel for accounting: Revenue model validationUse Claude to validate ASC 606 revenue models, surface reconciliation issues, and build financial visualizationsCategoryFinanceProductClaude.aiReading timeWatch time3min8minShareCopy linkhttps://claude.com/resources/tutorials/how-to-use-claude-in-excel-for-accounting-revenue-model-validationClaude in Excel works directly inside your spreadsheet through a sidebar, reading your data and making changes through conversation. In this tutorial, you'll see how an accountant uses Claude to validate an ASC 606 revenue recognition model—the kind of multi-tab workbook you might inherit during an audit or acquire in a transaction.What you'll learnUnderstand workbooks you didn't build — Ask Claude to explain how tabs connect. It reads every sheet first, then maps the structure—how schedules tie together, where data flows, what feeds into what.Surface errors and inconsistencies — Claude catches reconciliation gaps, duplicate entries, and missing data across multiple sheets. Instead of clicking through tabs yourself, you see everything it found upfront—and choose what to tackle first.Fix issues with confirmation — When Claude finds a problem, it shows you what's wrong, explains the recommended fix, and waits for your go-ahead. You stay in control of what changes.Add columns and formatting through conversation — Describe what you need—status indicators, legends, new calculations—and Claude builds it. It learns patterns from your existing data and applies them consistently.Create charts by describing them — Tell Claude what visualization you need. It creates helper tables when necessary, builds the chart, and will rebuild if the first version isn't right.In this videoFollow along with the video, or copy these prompts to try in your own workbook.0:00 — Get oriented in an unfamiliar modelAsk Claude to explain a workbook you've inherited or need to audit. Claude reads every tab before responding, giving it full context for follow-up questions."Walk me through this model. What's on each tab and how does everything tie together?"1:13 — Let Claude surface issues for youAfter reading the workbook, Claude proactively identifies problems: reconciliation gaps, duplicate entries, missing data. You choose which to tackle first."Yes, please look at the reconciliation issue first."2:17 — Fix data with confirmationClaude shows you what it found and what it recommends. It won't make changes until you say so."Yes, go ahead."3:09 — Add columns through conversationDescribe what you want—status indicators, legends, new calculations—and Claude adds them."Add a check mark column showing testing status for each contract. Use ✓ for tested, IP for in progress, and NP for not yet started. Add a legend at the bottom."4:24 — Create charts by describing themTell Claude what visualization you need. It creates helper tables when necessary and will rebuild if the first version isn't right."Create a waterfall chart showing the Q3 deferred revenue rollforward—beginning balance, bookings, revenue recognized, adjustments, and ending balance."6:11 — Return to earlier issuesClaude remembers everything from your conversation. Circle back to issues it flagged earlier without re-explaining."Let's go back to the issues you identified—a missing journal entry and one other I think?"What to noticeClaude reads before it answers. When you ask about the model, Claude doesn't guess—it reads all six sheets first, then explains how they connect. This prevents errors from incomplete context and gives Claude the baseline it needs for follow-up questions.Issues surface together. Instead of discovering problems one at a time as you click through tabs, Claude presents everything it found upfront: the reconciliation gap, the duplicate, the missing entry, the flagged items. You decide what to tackle first.Changes require confirmation. Claude shows the math, explains what will change, and waits for your "yes, go ahead" before modifying anything. You see exactly what's happening and can verify the logic.Context carries through the conversation. When you say "let's go back to the issues you identified," Claude remembers what it flagged earlier. You don't need to re-explain or re-read the model.Claude learns patterns from your data. When adding September journal entries, Claude analyzed the existing July and August entries, identified that revenue splits 80% Platform / 12% Support / 8% Services, and applied that same logic. The new entries follow your established pattern.Next stepsThis tutorial shows one use case: validating an ASC 606 revenue model with reconciliation issues and missing entries. The same approach works for any multi-sheet model where you need to understand the structure, catch problems, or extend what's already built—budgets, forecasts, audit workpapers.Get started:Install Claude in Excel from Microsoft AppSourceOpen it with Ctrl+Option+C (Mac) or Ctrl+Alt+C (Windows)Start with "Walk me through this model" on any spreadsheet you've inheritedLearn more:Claude in Excel help articleWhat is the Max plan?What you'll learnIn this videoWhat to noticeNext stepsRelated tutorialsGetting started with Claude in ExcelGetting started with Claude in ExcelGetting started with Claude in ExcelTutorialTutorialTutorialHow to use Claude in Excel for HR: Headcount planningHow to use Claude in Excel for HR: Headcount planningHow to use Claude in Excel for HR: Headcount planningTutorialTutorialTutorialPrompting strategies for financial analysisPrompting strategies for financial analysisPrompting strategies for financial analysisTutorialTutorialTutorialClaude for financial services overviewClaude for financial services overviewClaude for financial services overviewTutorialTutorialTutorial @container (width < 52em) { .footer_layout { display: flex; column-gap: var(--_spacing---space--4rem); row-gap: var(--_spacing---space--4rem); } .footer_content_wrap.u-column-custom { display: contents; } } .footer_link { opacity: 1; transition: opacity 0.3s ease; } .footer_links_list:has(.footer_link:hover) .footer_link:not(:hover) { opacity: 0.4; } HomepageHomepageNextNextThank you! Your submission has been received!Oops! Something went wrong while submitting the form. (function () { const ROOT_SEL = "[data-prompt-scope]"; const TRIGGER_SEL = "[data-prompt-trigger]"; const MENU_ATTR = "data-prompt-menu"; const CLOSE_SEL = "[data-prompt-menu-close]"; const ACTION_SEL = "a, button, [data-prompt-menu-action]"; const ITEM_SEL = "[data-prompt-item], li, .menu-item, .w-dyn-item"; // ---------- Claude helper ---------- function buildClaudeUrl(text) { const url = new URL("https://claude.ai/new"); url.searchParams.set("q", text || ""); return url.toString(); } function openClaude(text) { const q = (text || "").trim(); if (!q) return; window.open(buildClaudeUrl(q), "_blank", "noopener"); } // ---------- Utilities ---------- let uid = 0; function makeId(prefix = "prompt-menu") { uid += 1; return `${prefix}-${Date.now().toString(36)}-${uid}`; } // Limit queries to elements that belong to THIS scope (ignore nested scopes) function qsaInScope(root, sel) { return Array.from(root.querySelectorAll(sel)).filter( (el) => el.closest(ROOT_SEL) === root ); } // Lightweight shield to stop click-through during the brief close animation function deployClickShield(ms = 300) { const sh = document.createElement("div"); sh.style.cssText = "position:fixed;inset:0;z-index:2147483647;pointer-events:auto;background:transparent"; document.body.appendChild(sh); setTimeout(() => { sh.remove(); }, ms); } // Pairing: find which menu a trigger controls (no manual ids needed) function resolveMenuForTrigger(root, trigger) { // 1) explicit data-prompt-trigger="x" -> [data-prompt-menu="x"] const explicit = trigger.getAttribute("data-prompt-trigger"); if (explicit) { const m = qsaInScope(root, `[${MENU_ATTR}="${explicit}"]`)[0]; if (m) return m; } // 2) aria-controls const ctrl = trigger.getAttribute("aria-controls"); if (ctrl) { const m = qsaInScope(root, `#${ctrl.replace(/(["'\\])/g, "\\$1")}`)[0]; if (m) return m; } // 3) nearest following sibling with [data-prompt-menu] let sib = trigger.nextElementSibling; while (sib && sib !== root) { if ( sib.hasAttribute && sib.hasAttribute(MENU_ATTR) && sib.closest(ROOT_SEL) === root ) return sib; sib = sib.nextElementSibling; } // 4) first menu in this root return qsaInScope(root, `[${MENU_ATTR}]`)[0] || null; } function ensurePairing(trigger, menu) { if (!menu.id) menu.id = makeId(); trigger.setAttribute("aria-controls", menu.id); if (trigger.tagName === "BUTTON" && !trigger.hasAttribute("type")) trigger.type = "button"; } // ---------- Animations (GSAP) ---------- function revealMenu(menu) { return new Promise((resolve) => { gsap.set(menu, { visibility: "visible", pointerEvents: "auto", willChange: "transform, opacity", }); gsap.killTweensOf(menu); gsap.fromTo( menu, { opacity: 0, scale: 0.96 }, { opacity: 1, scale: 1, duration: 0.28, ease: "power3.out", clearProps: "willChange", onComplete: resolve, } ); }); } function hideMenu(menu) { return new Promise((resolve) => { gsap.killTweensOf(menu); gsap.to(menu, { opacity: 0, duration: 0.2, ease: "power2.out", onComplete: () => { gsap.set(menu, { visibility: "hidden", pointerEvents: "none", clearProps: "opacity,scale,willChange", }); resolve(); }, }); }); } // ---------- Per-scope controller ---------- const stateMap = new WeakMap(); // root -> { openMenuEl, openTrigger, isAnimating } function getState(root) { let s = stateMap.get(root); if (!s) { s = { openMenuEl: null, openTrigger: null, isAnimating: false }; stateMap.set(root, s); } return s; } function setTriggersInteractive(root, enabled) { qsaInScope(root, TRIGGER_SEL).forEach((el) => { el.style.pointerEvents = enabled ? "auto" : "none"; }); } // ---------- Button fade animations ---------- function fadeOutButtons(root) { return new Promise((resolve) => { const buttons = qsaInScope(root, TRIGGER_SEL); if (buttons.length === 0) { resolve(); return; } gsap.killTweensOf(buttons); gsap.to(buttons, { autoAlpha: 0, duration: 0.2, ease: "power2.out", onComplete: resolve, }); }); } function fadeInButtons(root) { return new Promise((resolve) => { const buttons = qsaInScope(root, TRIGGER_SEL); if (buttons.length === 0) { resolve(); return; } gsap.killTweensOf(buttons); gsap.to(buttons, { autoAlpha: 1, duration: 0.2, ease: "power2.out", onComplete: resolve, }); }); } async function openMenuIn(root, menu, trigger) { const s = getState(root); if (s.isAnimating || s.openMenuEl === menu) return; s.isAnimating = true; // close any currently open in THIS scope only if (s.openMenuEl && s.openMenuEl !== menu) { await hideMenu(s.openMenuEl); if (s.openTrigger) s.openTrigger.setAttribute("aria-expanded", "false"); } // Fade out buttons when opening menu await fadeOutButtons(root); await revealMenu(menu); s.openMenuEl = menu; s.openTrigger = trigger || null; if (s.openTrigger) s.openTrigger.setAttribute("aria-expanded", "true"); // Disable only this scope's triggers while open setTriggersInteractive(root, false); s.isAnimating = false; } async function closeMenuIn(root) { const s = getState(root); if (!s.openMenuEl || s.isAnimating) return; s.isAnimating = true; deployClickShield(280); // prevent click-through during closing await hideMenu(s.openMenuEl); if (s.openTrigger) s.openTrigger.setAttribute("aria-expanded", "false"); s.openMenuEl = null; s.openTrigger = null; // Re-enable only this scope's triggers setTriggersInteractive(root, true); // Fade in buttons when closing menu await fadeInButtons(root); s.isAnimating = false; } // ---------- Initializer ---------- const initedRoots = new WeakSet(); function initScope(root) { if (!root || initedRoots.has(root)) return; initedRoots.add(root); if (!window.gsap) { console.warn("GSAP is required for prompt menus."); return; } // Normalize ALL menus in this scope to hidden on init (prevents “open on load” bugs) qsaInScope(root, `[${MENU_ATTR}]`).forEach((menu) => { if (menu.dataset.pmMenuInit) return; menu.dataset.pmMenuInit = "true"; gsap.set(menu, { visibility: "hidden", opacity: 0, scale: 1, pointerEvents: "none", }); // Optional close button const closer = menu.querySelector(CLOSE_SEL); if (closer) { if (closer.tagName === "BUTTON" && !closer.hasAttribute("type")) closer.type = "button"; closer.addEventListener("click", (e) => { e.preventDefault(); e.stopPropagation(); closeMenuIn(root); }); } // Menu item -> Claude menu.addEventListener("click", (e) => { const target = e.target.closest(ACTION_SEL); if (!target || !menu.contains(target)) return; const container = target.closest(ITEM_SEL) || target; const hiddenP = container.querySelector("[data-prompt-menu-text]"); if (!hiddenP) return; e.preventDefault(); e.stopPropagation(); const text = hiddenP.textContent || hiddenP.innerText || ""; openClaude(text); // close after action closeMenuIn(root); }); }); // Bind triggers (auto-wire) qsaInScope(root, TRIGGER_SEL).forEach((trigger) => { if (trigger.dataset.pmTrigInit) return; trigger.dataset.pmTrigInit = "true"; const menu = resolveMenuForTrigger(root, trigger); if (menu) ensurePairing(trigger, menu); trigger.setAttribute("aria-expanded", "false"); trigger.addEventListener("click", async (e) => { e.preventDefault(); const targetMenu = resolveMenuForTrigger(root, trigger); if (!targetMenu) return; const s = getState(root); if (s.openMenuEl === targetMenu) { await closeMenuIn(root); // toggle close return; } await openMenuIn(root, targetMenu, trigger); }); }); // Outside click (this scope only) document.addEventListener("click", (e) => { const s = getState(root); if (!s.openMenuEl) return; const insideMenu = s.openMenuEl.contains(e.target); const onTrigger = !!e.target.closest(TRIGGER_SEL) && root.contains(e.target); if (!insideMenu && !onTrigger) closeMenuIn(root); }); // ESC (this scope only) document.addEventListener("keydown", (e) => { if (e.key === "Escape") closeMenuIn(root); }); } // ---------- Boot + observe ---------- function boot(container = document) { const roots = container.querySelectorAll ? container.querySelectorAll(ROOT_SEL) : []; roots.forEach(initScope); } if (document.readyState === "loading") { document.addEventListener("DOMContentLoaded", () => boot(), { once: true, }); } else { boot(); } const mo = new MutationObserver((muts) => { for (const m of muts) { for (const node of m.addedNodes || []) { if (node.nodeType !== 1) continue; if (node.matches && node.matches(ROOT_SEL)) initScope(node); if (node.querySelectorAll) { node.querySelectorAll(ROOT_SEL).forEach(initScope); } } } }); mo.observe(document.documentElement, { childList: true, subtree: true }); })(); (function () { // ---------- Claude helper ---------- function buildClaudeUrl(text) { const url = new URL('https://claude.ai/new'); url.searchParams.set('q', text || ''); return url.toString(); } // ---------- Auto-grow helpers ---------- function sizeTextarea(el) { // Ensure UA defaults don't force a starting height el.setAttribute('rows', el.getAttribute('data-min-rows') || '1'); el.style.minHeight = '0px'; el.style.height = 'auto'; // allow shrink el.style.height = el.scrollHeight + 'px'; // fit content } function initAutogrow(root = document) { root.querySelectorAll('textarea[data-autogrow]').forEach((el) => { if (el.dataset.autogrowInit) return; el.dataset.autogrowInit = 'true'; // Respect your existing pattern for default text const preset = el.getAttribute('data-default-text'); if (preset != null && !el.value) el.value = preset; // Initial sizing sizeTextarea(el); // Resize as you type el.addEventListener('input', () => sizeTextarea(el)); // In case fonts/styles load late and change line-height window.addEventListener('load', () => sizeTextarea(el), { once: true }); }); } // ---------- Claude form initializer ---------- function initClaudeForms(root = document) { root.querySelectorAll('form[data-claude-form]').forEach((form) => { if (form.dataset.claudeInit) return; form.dataset.claudeInit = 'true'; const textarea = form.querySelector('[data-claude-textarea], textarea'); const trigger = form.querySelector('[data-claude-button], button[type="button"], a[data-claude-button]'); if (!textarea || !trigger) { console.warn('Claude form: missing textarea or trigger in', form); return; } // If the trigger is a `; document.body.appendChild(el); return el; } const bubble = ensureBubble(); const elH = bubble.querySelector("#tt-title"); const elB = bubble.querySelector("#tt-body"); const elClose = bubble.querySelector(".tt-close"); // ---------------- Parse [[term|heading|body]] anywhere ---------------- const TOKEN_RE = /\[\[([^|\]]+)\|([^|\]]+)\|([^\]]+)\]\]/g; const BLOCK_SKIP = new Set(["SCRIPT","STYLE","NOSCRIPT","TEXTAREA","INPUT","SELECT","CODE","PRE","TEMPLATE","IFRAME"]); function shouldSkipTextNode(n){ let el = n.parentElement; while (el){ if (BLOCK_SKIP.has(el.tagName) || el.isContentEditable) return true; el = el.parentElement; } return false; } const walker = document.createTreeWalker(document.body, NodeFilter.SHOW_TEXT); const textNodes = []; while (walker.nextNode()){ const n = walker.currentNode; if (!n.nodeValue || shouldSkipTextNode(n)) continue; if (TOKEN_RE.test(n.nodeValue)) textNodes.push(n); TOKEN_RE.lastIndex = 0; } textNodes.forEach(node => { const frag = document.createDocumentFragment(); const insideLink = !!node.parentElement.closest("a"); let text = node.nodeValue, last = 0; TOKEN_RE.lastIndex = 0; let m; while ((m = TOKEN_RE.exec(text))){ if (m.index > last) frag.appendChild(document.createTextNode(text.slice(last, m.index))); const term=m[1].trim(), heading=m[2].trim(), body=m[3].trim(); const t = insideLink ? document.createElement("span") : document.createElement("button"); if (insideLink){ t.setAttribute("role","button"); t.setAttribute("tabindex","0"); } else { t.type="button"; } t.className="tt-trigger"; t.textContent=term; t.setAttribute("data-tt-h", heading); t.setAttribute("data-tt-b", body); t.setAttribute("aria-haspopup","dialog"); t.setAttribute("aria-expanded","false"); frag.appendChild(t); last = TOKEN_RE.lastIndex; } if (last < text.length) frag.appendChild(document.createTextNode(text.slice(last))); node.parentNode.replaceChild(frag, node); }); // ---------------- State ---------------- let current = null; let hoverCount = 0; let closeTimer = null; // Dimming bookkeeping let dimCtx = null; // { container, dimEls:[], wrappedTexts:[], pathEls:[] } // ---------------- Find the correct "text element" container ---------------- function findTextContainer(trigger){ // Prefer common RTE wrappers let el = trigger.closest(".w-richtext, .rich-text, .rte, [data-rte]"); if (el) return el; // Otherwise climb until we find an ancestor that contains multiple block nodes anywhere inside. const BLOCK_SEL = "p,h1,h2,h3,h4,h5,h6,ul,ol,li,blockquote,pre,figure,figcaption"; el = trigger.parentElement; while (el && el !== document.body){ const blockCount = el.querySelectorAll(BLOCK_SEL).length; if (blockCount >= 2) return el; el = el.parentElement; } // Fallback: nearest non-inline container el = trigger.parentElement || document.body; while (el && el !== document.body){ const d = getComputedStyle(el).display; if (d !== "inline" && d !== "contents") return el; el = el.parentElement; } return document.body; } // Utility: child of `ancestor` that contains `target` (direct child) function directChildContaining(ancestor, target){ for (const ch of ancestor.children){ if (ch === target || ch.contains(target)) return ch; } return null; } function getElementTarget(e) { // If target is already an Element, use it if (e.target instanceof Element) return e.target; // Otherwise, walk the composed/path for the first Element const path = (typeof e.composedPath === 'function') ? e.composedPath() : []; for (const n of path) if (n instanceof Element) return n; return null; } // ---------------- Dim everything except the trigger branch (sibling branches only) ---------------- function dimAllOtherBranches(container, trigger){ undim(); // clear previous const dimEls = []; const wrappedTexts = []; const pathEls = []; // Build ELEMENT-only path [container -> ... -> trigger] const path = []; for (let el = trigger; el && el !== container; el = el.parentElement) path.push(el); path.push(container); path.reverse(); // At each ancestor level, find the *direct* child that leads to the trigger for (let i = 0; i < path.length; i++){ const anc = path[i]; const branchChild = (i < path.length - 1) ? directChildContaining(anc, path[i+1]) : path[i]; // last step is the trigger itself // Fade element siblings (whole branches) for (const child of anc.children){ if (child === branchChild) continue; // keep the path branch crisp // Never fade any element that is (or contains) the trigger if (child === trigger || child.contains(trigger)) continue; child.style.transition = `opacity ${DIM_EASE_MS}ms ease`; child.style.opacity = String(DIM_OPACITY); dimEls.push(child); } // Fade TEXT NODE siblings directly under this ancestor (outside branchChild) anc.childNodes.forEach(node => { if (node.nodeType !== 3) return; // text only if (!node.nodeValue || !node.nodeValue.trim()) return; // If this text node sits inside branchChild, skip if (branchChild && branchChild.contains && branchChild.contains(node)) return; const span = document.createElement("span"); span.style.transition = `opacity ${DIM_EASE_MS}ms ease`; span.style.opacity = String(DIM_OPACITY); span.textContent = node.nodeValue; node.parentNode.replaceChild(span, node); wrappedTexts.push(span); }); // Keep a reference to the path elements (so we can explicitly restore opacity if needed) if (anc && anc.nodeType === 1) pathEls.push(anc); } // Hard-guard: explicitly set opacity:1 on the entire path to neutralize any inherited fade pathEls.forEach(el => { el.style.opacity = "1"; }); dimCtx = { container, dimEls, wrappedTexts, pathEls }; } function undim(){ if (!dimCtx) return; const { dimEls, wrappedTexts, pathEls } = dimCtx; // Animate back dimEls.forEach(el => { el.style.transition = `opacity ${DIM_EASE_MS}ms ease`; el.style.opacity = "1"; // remove inline style after the animation so we don't override site CSS setTimeout(() => { if (el) el.style.opacity = ""; }, DIM_EASE_MS + 50); }); wrappedTexts.forEach(span => { span.style.transition = `opacity ${DIM_EASE_MS}ms ease`; span.style.opacity = "1"; span.addEventListener("transitionend", () => { if (!span.parentNode) return; span.parentNode.replaceChild(document.createTextNode(span.textContent || ""), span); }, { once:true }); }); // Clear hard-guard on path pathEls.forEach(el => { if (el) el.style.opacity = ""; }); dimCtx = null; } // ---------------- Positioning (centered, edge-aware, flip) ---------------- function clamp(v,min,max){ return Math.max(min,Math.min(max,v)); } function measureBubbleForPlacement(){ const wasOpen = bubble.classList.contains("is-open"); if (!wasOpen){ bubble.style.visibility="hidden"; bubble.classList.add("is-open"); } const rect = bubble.getBoundingClientRect(); if (!wasOpen){ bubble.classList.remove("is-open"); bubble.style.visibility=""; } return { w: rect.width, h: rect.height }; } function placeAnchored(trigger){ const vw=innerWidth, vh=innerHeight; const r = trigger.getBoundingClientRect(); const { w, h } = measureBubbleForPlacement(); let left = r.left + (r.width/2) - (w/2); left = clamp(left, EDGE_PADDING, Math.max(EDGE_PADDING, vw - EDGE_PADDING - w)); const topBelow = r.bottom + OFFSET_Y; const spaceBelow = vh - topBelow - EDGE_PADDING; const placeBelow = spaceBelow >= h; let top = placeBelow ? topBelow : (r.top - h - OFFSET_Y); top = clamp(top, EDGE_PADDING, Math.max(EDGE_PADDING, vh - EDGE_PADDING - h)); bubble.style.left = left + "px"; bubble.style.top = top + "px"; const br = bubble.getBoundingClientRect(); if (br.bottom > vh - EDGE_PADDING){ bubble.style.maxHeight = (vh - 2*EDGE_PADDING) + "px"; bubble.style.overflowY = "auto"; } else { bubble.style.maxHeight = "none"; bubble.style.overflowY = "visible"; } } // ---------------- Open / Close (place → fade/scale) ---------------- function animateIn(){ bubble.style.transition = "none"; bubble.style.opacity = "0"; bubble.style.transform = "scale(0.95)"; void bubble.offsetWidth; bubble.style.transition = "opacity .18s ease, transform .18s ease"; bubble.style.opacity = "1"; bubble.style.transform = "scale(1)"; } function animateOut(done){ bubble.style.transition = "opacity .16s ease, transform .16s ease"; bubble.style.opacity = "0"; bubble.style.transform = "scale(0.95)"; const end = () => { bubble.removeEventListener("transitionend", end); done && done(); }; bubble.addEventListener("transitionend", end); setTimeout(end, 260); } function openFromTrigger(trigger){ if (current && current !== trigger) forceClose(); current = trigger; trigger.setAttribute("aria-expanded","true"); elH.textContent = trigger.getAttribute("data-tt-h") || ""; elB.textContent = trigger.getAttribute("data-tt-b") || ""; bubble.classList.add("is-open"); bubble.setAttribute("aria-hidden","false"); placeAnchored(trigger); animateIn(); const container = findTextContainer(trigger); dimAllOtherBranches(container, trigger); hoverCount = 0; cancelCloseTimer(); } function forceClose(){ if (!current) return; bubble.classList.remove("is-open"); bubble.setAttribute("aria-hidden","true"); current.setAttribute("aria-expanded","false"); current = null; undim(); hoverCount = 0; cancelCloseTimer(); } function closeWithAnim(){ if (!current) return; const t = current; animateOut(() => { bubble.classList.remove("is-open"); bubble.setAttribute("aria-hidden","true"); t.setAttribute("aria-expanded","false"); current = null; undim(); }); } function scheduleClose(){ cancelCloseTimer(); closeTimer = setTimeout(() => { if (hoverCount <= 0 && !isCoarse()) closeWithAnim(); }, CLOSE_DELAY); } function cancelCloseTimer(){ if (closeTimer){ clearTimeout(closeTimer); closeTimer = null; } } // ---------------- Hover-intent (desktop) ---------------- function onZoneEnter(){ if (isCoarse()) return; hoverCount++; cancelCloseTimer(); } function onZoneLeave(){ if (isCoarse()) return; hoverCount = Math.max(0, hoverCount - 1); if (hoverCount === 0) scheduleClose(); } bubble.addEventListener("pointerenter", onZoneEnter, true); bubble.addEventListener("mouseenter", onZoneEnter, true); bubble.addEventListener("pointerleave", onZoneLeave, true); bubble.addEventListener("mouseleave", onZoneLeave, true); const handleEnter = (e) => { if (isCoarse()) return; const target = getElementTarget(e); if (!target) return; const t = target.closest(".tt-trigger"); if (!t) return; onZoneEnter(); if (!current || current !== t) openFromTrigger(t); }; const handleLeave = (e) => { if (isCoarse()) return; const target = getElementTarget(e); if (!target) return; const t = target.closest(".tt-trigger"); if (!t) return; onZoneLeave(); }; document.addEventListener("pointerenter", handleEnter, true); document.addEventListener("mouseenter", handleEnter, true); document.addEventListener("pointerleave", handleLeave, true); document.addEventListener("mouseleave", handleLeave, true); // ---------------- Keyboard ---------------- document.addEventListener("focusin", (e) => { if (!e.target) return; const t = e.target.closest(".tt-trigger"); if (t) openFromTrigger(t); }); document.addEventListener("focusout", (e) => { if (!e.target) return; const t = e.target.closest(".tt-trigger"); if (t && current === t) closeWithAnim(); }); // ---------------- Mobile / coarse ---------------- document.addEventListener("pointerdown", (e) => { if (!isCoarse()) return; const t = e.target.closest(".tt-trigger"); if (!t) return; e.preventDefault(); e.stopPropagation(); if (current === t && bubble.classList.contains("is-open")) { closeWithAnim(); return; } openFromTrigger(t); }, true); document.addEventListener("click", (e) => { if (!isCoarse()) return; if (!bubble.classList.contains("is-open")) return; const inBubble = !!e.target.closest(".tt-bubble"); const onTrigger = !!e.target.closest(".tt-trigger"); if (!inBubble && !onTrigger) closeWithAnim(); }, true); // Close button + ESC elClose.addEventListener("click", closeWithAnim); document.addEventListener("keydown", (e) => { if (e.key === "Escape") closeWithAnim(); }); // Reposition on resize/scroll while open const reposition = () => { if (!current) return; placeAnchored(current); }; addEventListener("resize", reposition, { passive: true }); addEventListener("scroll", reposition, { passive: true }); }); /* Tooltip Styles */ /* Trigger */ .tt-trigger { cursor: help; text-decoration: underline dotted; text-underline-offset: .2em; color: inherit; } .tt-trigger:focus-visible{ outline:2px solid currentColor; outline-offset:2px; } /* Bubble */ .tt-bubble{ position: fixed; z-index: 10; max-width: 17rem; background: var(--_theme---background-primary); box-shadow: 0 4px 24px rgba(0,0,0,.05); border-radius: var(--radius--large); border-style: solid; border-color: var(--_theme---border-tertiary); padding: var(--_spacing---space--1-5rem); pointer-events: none; opacity: 0; transform: translate3d(0,0,0) scale(.98); transition: opacity .3s ease, transform .3s ease; will-change: transform, opacity; } .tt-bubble.is-open{ opacity:1; transform:translate3d(0,0,0) scale(1); pointer-events:auto; } .tt-h{ margin-bottom: var(--_spacing---space--0-5rem); font-size: var(--_typography---font-size--body-3); font-family: var(--_typography---font--secondary-family); line-height: var(--_typography---line-height--1-6); color: var(--_theme---foreground-primary); } .tt-b{ margin:0; font-size: var(--_typography---font-size--caption); line-height: var(--_typography---line-height--1-6); color: var(--_theme---foreground-tertiary); } /* Mobile close button */ .tt-close { display: none; } @media (hover: none), (pointer: coarse) { .tt-close { display: inline-flex; position: absolute; top: 0.5rem; right: 0.5rem; width: 32px; height: 32px; align-items: center; justify-content: center; border: 0; border-radius: 999px; background: transparent; font-size: 22px; line-height: 1; color: inherit; cursor: pointer; touch-action: manipulation; } .tt-close:focus-visible { outline: 2px solid currentColor; outline-offset: 2px; } .tt-close:hover { opacity: 1; } } html[lang="de-DE"] h1, html[lang="de-DE"] h2, html[lang="de-DE"] h3, html[lang="de-DE"] h4, html[lang="de-DE"] h5, html[lang="de-DE"] h6, html[lang="de-DE"] p, html[lang="de-DE"] li, html[lang="fr-FR"] h1, html[lang="fr-FR"] h2, html[lang="fr-FR"] h3, html[lang="fr-FR"] h4, html[lang="fr-FR"] h5, html[lang="fr-FR"] h6, html[lang="fr-FR"] p, html[lang="fr-FR"] li { overflow-wrap: break-word; hyphens: auto; } :root { --nav--icon-thickness: var(--border-width--main); --nav--hamburger-thickness: var(--nav--icon-thickness); --nav--hamburger-gap: var(--_spacing---space--0-25rem); --nav--hamburger-rotate: 45; --nav--dropdown-duration: 300ms; --nav--dropdown-open-duration: 600ms; --nav--dropdown-delay: 0ms; --ease-expo-out: cubic-bezier(0.16, 1, 0.3, 1); } /* ========== GENERAL RESPONSIVE RULES ========== */ /* Lock body when nav is open (script toggles .is-nav-open) */ @media (width < 56em) { body.is-nav-open { overflow: hidden; } } @container (min-width: 56em) { .nav_wrap.is-desktop { display: block; } .nav_wrap.is-mobile { display: none; } } @container (width < 28em) { .nav_mobile_layout .nav_actions_mobile { display: none; } } /* ========== DROPDOWN STYLING ========== */ html:not(.wf-design-mode) .nav_dropdown_component > .w-dropdown-list { display: grid !important; grid-template-columns: minmax(0, 1fr); grid-template-rows: 0fr; transition: grid-template-rows var(--nav--dropdown-duration) var(--ease-expo-out), visibility 0s var(--nav--dropdown-duration), opacity var(--nav--dropdown-duration) var(--ease-expo-out); visibility: hidden; opacity: 0; } html:not(.wf-design-mode) .nav_dropdown_component > .w-dropdown-list.w--open { visibility: visible; opacity: 1; transition: grid-template-rows var(--nav--dropdown-duration) var(--ease-expo-out), visibility 0s 0s, opacity var(--nav--dropdown-duration) var(--ease-expo-out); } .nav_dropdown_component > .w-dropdown-list > * { overflow: hidden; } .nav_dropdown_component:has(> .w-dropdown-toggle[aria-expanded="true"]) > .w-dropdown-list { --nav--dropdown-duration: var(--nav--dropdown-open-duration); grid-template-rows: 1fr; } /*.nav_wrap.is-desktop:has(.nav_dropdown_component > .w-dropdown-toggle.w--open[aria-expanded="false"]) .nav_dropdown_component:has(> .w--open[aria-expanded="true"]) > .w-dropdown-list { transition-delay: var(--nav--dropdown-duration); }*/ /* Dropdown caret rotation */ .nav_links_svg.is-desktop { transition: transform 750ms var(--ease-expo-out); } .w-dropdown-toggle[aria-expanded="true"] .nav_links_svg.is-desktop { transform: rotate(-180deg); } .nav_links_svg_line.is-2 { transition: transform 500ms var(--ease-expo-out); } .w-dropdown-toggle[aria-expanded="true"] .nav_links_svg_line.is-2 { transform: rotate(0deg); } /* open (replicates your original transforms) */ .nav_btn_wrap[aria-expanded="true"] .nav_btn_line:nth-child(1), .nav_btn_wrap[aria-expanded="true"] > * > :first-child { transform: translateY(calc(var(--nav--hamburger-thickness) * 1 + var(--nav--hamburger-gap) * 1)) rotate(calc(var(--nav--hamburger-rotate) * -1deg)); } .nav_btn_wrap[aria-expanded="true"] .nav_btn_line:nth-child(2), .nav_btn_wrap[aria-expanded="true"] > * > :nth-child(2) { opacity: 0; } .nav_btn_wrap[aria-expanded="true"] .nav_btn_line:nth-child(3), .nav_btn_wrap[aria-expanded="true"] > * > :last-child { transform: translateY(calc(var(--nav--hamburger-thickness) * -1 + var(--nav--hamburger-gap) * -1)) rotate(calc(var(--nav--hamburger-rotate) * 1deg)); width: 1rem; } /* ========== HOVER & THEME EFFECTS ========== */ @media (hover: hover) and (pointer: fine) { body:has(.nav_dropdown_item:hover) .nav_dropdown_item:not(:hover) > * > * { color: var(--_theme---foreground-tertiary); } .nav_dropdown_link { transition: background-color 300ms ease, color 300ms ease; } .nav_dropdown_item:hover .nav_dropdown_link { background: var(--_theme---background-tertiary); color: var(--_theme---foreground-primary); } .nav_secondary_wrap .nav_dropdown_item:hover .nav_dropdown_link { background: var(--_theme---background-tertiary); color: var(--_theme---foreground-primary); } .nav_wrap.is-mobile .nav_dropdown_item:hover .nav_dropdown_link { color: var(--_theme---foreground-primary); } .nav_links_text { transition: color 500ms var(--ease-expo-out); } .nav_links_svg { transition: transform 500ms var(--ease-expo-out), color 500ms var(--ease-expo-out); } .nav_links_item:hover .nav_links_text, .nav_links_item:hover .nav_links_svg { color: var(--_theme---foreground-primary); } } /* ========== LAYOUT / UTILITY (kept, de-Webflow’d) ========== */ .nav_wrap.is-mobile [data-open-on-mobile] > .w-dropdown-toggle { display: none; } .nav_wrap.is-mobile [data-open-on-mobile] > .w-dropdown-list { visibility: visible; opacity: 1; display: block; grid-template-rows: 1fr; } .nav_buttons_item .button_main_wrap { width: 100%; min-width: max-content; } /* Optional: fade out mobile actions while open */ .nav_actions_wrap { transition: opacity 500ms var(--ease-expo-out); } body.is-nav-open .nav_actions_mobile .nav_actions_wrap { opacity: 0; pointer-events: none; } .nav_links_item:first-child { border-top: none; } /* ========== Breadcrumbs ========== */ .breadcrumb_text.is_linked[href="#"], .breadcrumb_text:has(+.breadcrumb_text.is_linked:not([href="#"])) { display: none; } .breadcrumb_text.is_linked:not([href="#"]) { display: block; } document.addEventListener("DOMContentLoaded", function () { document.querySelectorAll(".nav_component").forEach((root) => { if (root.dataset.scriptInitialized) return; root.dataset.scriptInitialized = "true"; if (!window.gsap) { console.error("GSAP not found"); return; } const btn = root.querySelector('.nav_btn_wrap'); const menu = root.querySelector('.nav_menu_wrap'); if (!btn || !menu) { console.warn('Missing .nav_btn_wrap or .nav_menu_wrap in', root); return; } // a11y setup (scoped) if (!btn.hasAttribute('type')) btn.setAttribute('type', 'button'); if (!btn.hasAttribute('aria-expanded')) btn.setAttribute('aria-expanded', 'false'); if (!menu.id) menu.id = 'primary-nav-' + Math.random().toString(36).slice(2); if (!btn.hasAttribute('aria-controls')) btn.setAttribute('aria-controls', menu.id); menu.setAttribute('aria-hidden', 'true'); // targets (scoped) const items = Array.from(menu.querySelectorAll('.nav_links_item')); const actions = menu.querySelector('.nav_menu_actions_wrap'); // feature detection const prefersReduced = window.matchMedia('(prefers-reduced-motion: reduce)').matches; const canClipInset = !!(window.CSS && CSS.supports && ( CSS.supports('clip-path','inset(0 0 100% 0)') || CSS.supports('-webkit-clip-path','inset(0 0 100% 0)') )); const useClip = canClipInset && !prefersReduced; // durations from CSS vars function readDur(varName, fallbackSec) { const v = getComputedStyle(document.documentElement).getPropertyValue(varName).trim(); if (!v) return fallbackSec; if (v.endsWith('ms')) return parseFloat(v)/1000; if (v.endsWith('s')) return parseFloat(v); const n = parseFloat(v); return isNaN(n) ? fallbackSec : n; } const OPEN_DUR = readDur('--nav--menu-open-duration', 0.8); const CLOSE_DUR = readDur('--nav--menu-close-duration', 0.4); // state per component let isOpen = false; let current = null; function setMenuVisibleForAnim() { menu.style.display = 'flex'; // ensure it's shown before anim menu.removeAttribute('hidden'); menu.setAttribute('aria-hidden', 'false'); menu.style.willChange = useClip ? 'clip-path' : 'transform, opacity'; } function clearMenuInline() { gsap.set(menu, { clearProps: 'clipPath,webkitClipPath,opacity,transform,willChange,pointerEvents' }); } // OPEN function playOpen() { setMenuVisibleForAnim(); menu.style.pointerEvents = 'none'; document.body.classList.add('is-nav-open'); btn.setAttribute('aria-expanded', 'true'); if (useClip) { gsap.set(menu, { clipPath: 'inset(0 0 100% 0)', webkitClipPath: 'inset(0 0 100% 0)' }); } else { gsap.set(menu, { yPercent: -2, opacity: 0 }); } if (items.length) gsap.set(items, { y: 20, autoAlpha: 0 }); if (actions) gsap.set(actions, { y: 20, autoAlpha: 0 }); const tl = gsap.timeline({ defaults: { ease: 'power3.out' } }); if (useClip) { tl.to(menu, { clipPath: 'inset(0 0 0% 0)', webkitClipPath: 'inset(0 0 0% 0)', duration: prefersReduced ? 0.01 : OPEN_DUR, ease: 'expo.out' }, 0); } else { tl.to(menu, { yPercent: 0, opacity: 1, duration: prefersReduced ? 0.01 : Math.min(OPEN_DUR, 0.36) }, 0.02); } if (items.length) { tl.to(items, { y: 0, autoAlpha: 1, stagger: prefersReduced ? 0 : 0.08, duration: prefersReduced ? 0.01 : 0.4 }, 0.10); } if (actions) { const base = 0.10 + (items.length ? items.length * (prefersReduced ? 0 : 0.08) : 0); tl.to(actions, { y: 0, autoAlpha: 1, duration: prefersReduced ? 0.01 : 0.4 }, base); } tl.add(() => { menu.style.pointerEvents = 'auto'; }, '>-0.1'); // listeners (per open) document.addEventListener('keydown', onKeydown); menu.addEventListener('click', onMenuLinkClick); return tl; } // CLOSE (fade all together, then clip inset close) function playClose() { menu.style.pointerEvents = 'none'; btn.setAttribute('aria-expanded', 'false'); document.body.classList.remove('is-nav-open'); const tl = gsap.timeline({ defaults: { ease: 'power3.out' } }); const fadeTargets = items.concat(actions ? [actions] : []); if (fadeTargets.length) tl.to(fadeTargets, { autoAlpha: 0, y: 0, duration: prefersReduced ? 0.01 : 0.2 }, 0); if (useClip) { gsap.set(menu, { clipPath: 'inset(0 0 0% 0)', webkitClipPath: 'inset(0 0 0% 0)' }); tl.to(menu, { clipPath: 'inset(0 0 100% 0)', webkitClipPath: 'inset(0 0 100% 0)', duration: prefersReduced ? 0.01 : CLOSE_DUR }, '>-0.02'); } else { tl.to(menu, { yPercent: -2, opacity: 0, duration: prefersReduced ? 0.01 : Math.min(CLOSE_DUR, 0.28) }, '>-0.02'); } tl.add(() => { menu.style.display = 'none'; menu.setAttribute('aria-hidden', 'true'); clearMenuInline(); if (items.length) gsap.set(items, { clearProps: 'all' }); if (actions) gsap.set(actions, { clearProps: 'all' }); // remove listeners added on open document.removeEventListener('keydown', onKeydown); menu.removeEventListener('click', onMenuLinkClick); }); return tl; } function openMenu() { if (isOpen) return; isOpen = true; if (current && current.isActive()) current.kill(); current = playOpen(); } function closeMenu() { if (!isOpen) return; isOpen = false; if (current && current.isActive()) current.kill(); current = playClose(); } function onKeydown(e){ if (e.key === 'Escape' && isOpen) { e.preventDefault(); closeMenu(); } } function onMenuLinkClick(e){ const a = e.target.closest('a[href]'); if (!a) return; const url = new URL(a.href, location.href); if (url.origin === location.origin) { e.preventDefault(); const tl = playClose(); tl.eventCallback('onComplete', () => { window.location.href = a.href; }); isOpen = false; } } // Toggle (scoped) btn.addEventListener('click', () => (isOpen ? closeMenu() : openMenu())); // Normalize if visible on load (scoped) if (getComputedStyle(menu).display !== 'none') { btn.setAttribute('aria-expanded', 'true'); menu.setAttribute('aria-hidden', 'false'); document.body.classList.add('is-nav-open'); isOpen = true; } }); }); (function () { 'use strict'; // ---------- tiny utils ---------- var NS = 'navBundleInit'; function onceFlag(el, k){ k=k||'scriptInitialized'; if (el.dataset[k]) return true; el.dataset[k]='true'; return false; } function ready(fn){ if(document.readyState==='loading'){document.addEventListener('DOMContentLoaded',fn,{once:true});} else { fn(); } } // ---------- 1) Ask Claude about this page ---------- function initAskPage(){ var buttons = document.querySelectorAll('[data-ask-page]'); if (!buttons.length) return; // Early exit if no buttons buttons.forEach(function(btn){ if (onceFlag(btn, NS)) return; btn.addEventListener('click', function (e) { e.preventDefault(); var pageUrl = window.location.href; var prompt = "Read this page " + pageUrl + " so that I can ask you questions about it"; var claudeUrl = new URL('https://claude.ai/new'); claudeUrl.searchParams.set('q', prompt); window.open(claudeUrl.toString(), '_blank', 'noopener'); }); }); } // ---------- 2) Copy page content as Markdown (Turndown) ---------- var _turndownReady; function ensureTurndown(){ if (window.TurndownService) return Promise.resolve(); if (_turndownReady) return _turndownReady; _turndownReady = new Promise(function(resolve, reject){ var s = document.createElement('script'); s.src = 'https://unpkg.com/turndown/dist/turndown.js'; s.async = true; s.onload = function(){ resolve(); }; s.onerror = function(){ reject(new Error('Failed to load Turndown')); }; document.head.appendChild(s); }); return _turndownReady; } function initCopyAsMarkdown(){ var copyButton = document.getElementById('copy-as-markdown'); if (!copyButton) return; // Early exit if (onceFlag(copyButton, NS)) return; var buttonTextEl = copyButton.querySelector('.nav_dropdown_text') || copyButton; var originalText = buttonTextEl.textContent; copyButton.addEventListener('click', function(){ ensureTurndown().then(function(){ try { var TurndownService = window.TurndownService; var turndownService = new TurndownService({ headingStyle: 'atx', codeBlockStyle: 'fenced', fence: '```', emDelimiter: '*', strongDelimiter: '**', linkStyle: 'inlined' }); // Skip junk turndownService.addRule('skipWebflowElements', { filter: function(node){ return node.nodeName === 'SCRIPT' || node.nodeName === 'STYLE' || (node.className && (String(node.className).includes('w-editor') || String(node.className).includes('w-embed'))); }, replacement: function(){ return ''; } }); buttonTextEl.textContent = 'Copying...'; copyButton.disabled = true; var contentElement = document.querySelector('main') || document.querySelector('.main-content') || document.querySelector('body'); if (!contentElement) throw new Error('No content found to copy'); var cloned = contentElement.cloneNode(true); cloned.querySelectorAll('script, style, nav, footer, .w-nav, .footer').forEach(function(el){ el.remove(); }); var markdown = turndownService.turndown(cloned); var done = function(){ buttonTextEl.textContent = 'Copied!'; setTimeout(function(){ buttonTextEl.textContent = originalText; copyButton.disabled = false; }, 2000); }; if (navigator.clipboard && window.isSecureContext) { navigator.clipboard.writeText(markdown).then(done, function(err){ throw err; }); } else { var ta = document.createElement('textarea'); ta.value = markdown; ta.style.position='fixed'; ta.style.opacity='0'; document.body.appendChild(ta); ta.select(); document.execCommand('copy'); document.body.removeChild(ta); done(); } } catch (err){ console.error('Copy failed:', err); buttonTextEl.textContent = 'Copy failed'; setTimeout(function(){ buttonTextEl.textContent = originalText; copyButton.disabled = false; }, 2000); } }).catch(function(err){ console.error('Turndown load failed:', err); buttonTextEl.textContent = 'Copy failed'; setTimeout(function(){ buttonTextEl.textContent = originalText; copyButton.disabled = false; }, 2000); }); }); } // ---------- init all ---------- ready(function(){ initAskPage(); initCopyAsMarkdown(); }); // Optional: expose minimal API for debugging (comment out in production if not needed) // window.NavBundle = { // initCopyAsMarkdown: initCopyAsMarkdown, // initAskPage: initAskPage // }; })(); Explore here /* - - - - - - - - - - - - - - - - - - - - - - - - - - - THUMBNAIL BG COLORS - - - - - - - - - - - - - - - - - - - - - - - - - - - - */ [data-thumbnail-bg="Clay"] { background-color: var(--swatch--clay); } [data-thumbnail-bg="Cactus"] { background-color: var(--swatch--cactus); } [data-thumbnail-bg="Coral"] { background-color: var(--swatch--coral); } [data-thumbnail-bg="Fig"] { background-color: var(--swatch--fig); } [data-thumbnail-bg="Heather"] { background-color: var(--swatch--heather); } [data-thumbnail-bg="Oat"] { background-color: var(--swatch--oat); } [data-thumbnail-bg="Olive"] { background-color: var(--swatch--olive); } [data-thumbnail-bg="Sky"] { background-color: var(--swatch--sky); } /* - - - - - - - - - - - - - - - - - - - - - - - - - - - CONDITIONAL STYLES BASED ON IF IMAGE OR VIDEO ARE VISIBLE - - - - - - - - - - - - - - - - - - - - - - - - - - - - */ .hero_tutorial_post_wrap:has(+ .tutorial_post_media_wrap :is(.tutorial_post_image_wrap, .tutorial_post_video_wrap):not(.w-condition-invisible)) { border-bottom: none !important; } .tutorial_post_media_wrap:has(+ .tutorial_post_wrap:not(.w-condition-invisible)) .u-section-spacer { display: none !important; } /* - - - - - - - - - - - - - - - - - - - - - - - - - - - TABLE OF CONTENTS Adding here so classes don't get accidentally cleared - - - - - - - - - - - - - - - - - - - - - - - - - - - - */ .tutorial_post_toc_list { color: var(--_theme---foreground-tertiary); font-size: var(--_typography---font-size--body-3); } .tutorial_post_toc_item { margin-left: -0.0625rem; padding-top: var(--_spacing---space--0-25rem); padding-bottom: var(--_spacing---space--0-25rem); padding-left: var(--_spacing---space--1rem); border-left-style: solid; border-left-width: var(--border-width--main); border-left-color: var(--_theme---border-secondary); } .tutorial_post_toc_item.active { border-left-color: var(--_theme---heroes-accent); } /* - - - - - - - - - - - - - - - - - - - - - - - - - - - ANCHOR LINK - - - - - - - - - - - - - - - - - - - - - - - - - - - - */ /* offset for that takes fixed header height into account */ .anchor { position: relative; top: calc(-1 * var(--_spacing---space--4rem) - 120px); } /* Copy anchor link to clipboard button */ .anchor-link-box { /* quick btn reset */ background: none; border: none; padding: 0; cursor: pointer; /* other styles */ position: absolute !important; left: -3rem; top: 50%; transform: translateY(-50%); opacity: 0; transition: opacity 0.2s ease; width: 2.5rem; height: 2.5rem; background: var(--_theme---background-tertiary); border: var(--border-width--main) solid var(--_theme---border-tertiary); border-radius: var(--radius--main); display: flex; align-items: center; justify-content: center; padding: var(--_spacing---space--0-5rem); } .anchor-link-box:hover { opacity: 1; } .tooltip { opacity: 0; transition: opacity 0.2s ease; position: absolute; left: 50%; bottom: calc(100% + 0.5rem); transform: translateX(-50%); background: var(--_theme---button-primary--background); color: var(--_theme---button-primary--text); white-space: nowrap; padding: 0.125rem 0.5rem; border-radius: var(--radius--x-small); font-family: var(--_typography---font--primary-family); font-size: var(--_typography---font-size--caption); line-height: var(--_typography---line-height--1-6); } .tooltip.active { opacity: 1; } /* - - - - - - - - - - - - - - - - - - - - - - - - - - - HIDE EMPTY

TAGS - - - - - - - - - - - - - - - - - - - - - - - - - - - - */ p:empty, p:has(> br:only-child) { display: none; } /* - - - - - - - - - - - - - - - - - - - - - - - - - - - BORDER RADIUS ON RICH TEXT VIDEOS - - - - - - - - - - - - - - - - - - - - - - - - - - - - */ .w-richtext-figure-type-video { border-radius: var(--radius--main); overflow: hidden; } /* - - - - - - - - - - - - - - - - - - - - - - - - - - - FIX GRID SPACING IF TUTORIAL HAS NO VIDEO - - - - - - - - - - - - - - - - - - - - - - - - - - - - */ .tutorial_post_video_layout:has(.w-dyn-bind-empty) { display: none; } document.addEventListener('DOMContentLoaded', function() { // Helper function to convert text to slug function textToSlug(text) { return text .toLowerCase() .trim() .replace(/[''"&]/g, '') .replace(/"([^"]+)"/, '$1') .replace(/\s*[:.,?—–\\/-]\s*/g, '-') .replace(/\s+/g, '-') .replace(/-+/g, '-') .replace(/^-+|-+$/g, ''); } // Track used slugs to prevent duplicate IDs const usedSlugs = new Set(); function getUniqueSlug(baseSlug) { let slug = baseSlug; let counter = 1; while (usedSlugs.has(slug)) { slug = `${baseSlug}-${counter}`; counter++; } usedSlugs.add(slug); return slug; } // Add/Remove class helper functions const addClass = (element, className) => { if (element && className) { element.classList.add(className); } }; const removeClass = (element, className) => { if (element && className) { element.classList.remove(className); } }; // Debounce helper for resize events function debounce(func, wait) { let timeout; return function executedFunction(...args) { const later = () => { clearTimeout(timeout); func(...args); }; clearTimeout(timeout); timeout = setTimeout(later, wait); }; } // Get elements const tutorialContent = document.getElementById('tutorial_content'); const tocContainer = document.getElementById('tutorial_toc'); if (!tutorialContent || !tocContainer) { return; } const h2Elements = tutorialContent.querySelectorAll('h2'); if (h2Elements.length === 0) { tocContainer.style.display = 'none'; return; } // Create ToC list const tocList = document.createElement('ul'); addClass(tocList, 'tutorial_post_toc_list'); const tocItems = []; // Copy anchor link to clipboard with tooltip const copyAnchorToClipboard = (element, spanElement, mouseEnterTime = 750) => { let hoverTimeout; let tooltip; element.addEventListener("mouseenter", function () { clearTimeout(hoverTimeout); tooltip = document.createElement("span"); addClass(tooltip, 'tooltip'); tooltip.innerText = "Copy anchor link"; this.appendChild(tooltip); hoverTimeout = setTimeout(() => { addClass(tooltip, "active"); }, mouseEnterTime); }); element.addEventListener("mouseleave", function () { clearTimeout(hoverTimeout); if (tooltip) { removeClass(tooltip, "active"); this.removeChild(tooltip); tooltip = null; } }); element.addEventListener("click", function (event) { event.preventDefault(); clearTimeout(hoverTimeout); if (spanElement && spanElement.classList.contains("anchor")) { const id = spanElement.id; if (tooltip) { tooltip.innerText = "Copied!"; addClass(tooltip, "active"); const url = new URL(window.location.href); url.searchParams.delete("topic"); const anchorLink = `${url.origin}${url.pathname}#${id}`; if (navigator && navigator.clipboard) { navigator.clipboard.writeText(anchorLink).catch((err) => { console.error("Could not copy text: ", err); }); } hoverTimeout = setTimeout(() => { removeClass(tooltip, "active"); }, 1500); } } else { console.error("Couldn't find Span with the ID"); } }); }; // Anchor link SVG icon const anchorLinkSVG = ``; // Process each H2 and build ToC h2Elements.forEach((h2) => { const headingText = h2.textContent; const baseSlug = textToSlug(headingText); const slug = getUniqueSlug(baseSlug); // Add anchor span const anchorSpan = document.createElement('span'); anchorSpan.id = slug; addClass(anchorSpan, 'anchor'); h2.prepend(anchorSpan); // Add anchor link box const anchorLinkBox = document.createElement('button'); anchorLinkBox.type = 'button'; addClass(anchorLinkBox, 'anchor-link-box'); anchorLinkBox.innerHTML = anchorLinkSVG; h2.prepend(anchorLinkBox); // Initialize copy functionality copyAnchorToClipboard(anchorLinkBox, anchorSpan); // Create ToC item const listItem = document.createElement('li'); addClass(listItem, 'tutorial_post_toc_item'); const link = document.createElement('a'); link.href = `#${slug}`; const textDiv = document.createElement('div'); textDiv.textContent = headingText; link.appendChild(textDiv); listItem.appendChild(link); tocList.appendChild(listItem); tocItems.push({ element: listItem, target: h2 }); }); tocContainer.appendChild(tocList); // Configuration const headerHeight = 134; const activationOffset = 100; // State let currentActive = 0; let ranges = []; let frameRequested = false; // Calculate activation ranges using getBoundingClientRect for reliability // (offsetTop can be relative to offsetParent, not document) function calculateRanges() { const scrollY = window.scrollY; ranges = tocItems.map((item, index) => { const rect = item.target.getBoundingClientRect(); const absoluteTop = rect.top + scrollY; const activationPoint = absoluteTop - headerHeight - activationOffset; const nextAbsoluteTop = index < tocItems.length - 1 ? tocItems[index + 1].target.getBoundingClientRect().top + scrollY : Infinity; const nextActivation = nextAbsoluteTop === Infinity ? Infinity : nextAbsoluteTop - headerHeight - activationOffset; return { index: index, start: Math.max(0, activationPoint), end: nextActivation }; }); } // Update active section function updateActiveSection() { if (ranges.length === 0) return; const scrollPosition = window.scrollY; let newActive = 0; // Find current range if (scrollPosition < ranges[0].start) { newActive = 0; } else { for (let i = 0; i < ranges.length; i++) { if (scrollPosition >= ranges[i].start && scrollPosition < ranges[i].end) { newActive = i; break; } } } // Special case: if near bottom of page, activate last item const scrollBottom = scrollPosition + window.innerHeight; const documentHeight = document.documentElement.scrollHeight; if (documentHeight - scrollBottom < 100) { newActive = tocItems.length - 1; } // Update classes if changed if (newActive !== currentActive) { tocItems.forEach(item => removeClass(item.element, 'active')); currentActive = newActive; if (tocItems[currentActive]) { addClass(tocItems[currentActive].element, 'active'); } } frameRequested = false; } // Scroll handler function handleScroll() { if (!frameRequested) { frameRequested = true; window.requestAnimationFrame(updateActiveSection); } } // Recalculate on resize (debounced) const handleResize = debounce(() => { calculateRanges(); updateActiveSection(); }, 250); // Smooth scroll behavior for ToC links tocItems.forEach(item => { item.element.querySelector('a').addEventListener('click', function(e) { e.preventDefault(); const targetId = this.getAttribute('href').substring(1); const targetElement = document.getElementById(targetId); if (targetElement) { const scrollTo = targetElement.parentElement.offsetTop - headerHeight - 20; window.scrollTo({ top: scrollTo, behavior: 'smooth' }); } }); }); // Initialize calculateRanges(); addClass(tocItems[0].element, 'active'); // Handle hash on initial page load if (window.location.hash) { const hash = window.location.hash.substring(1); const targetIndex = tocItems.findIndex(item => item.target.querySelector('.anchor')?.id === hash ); if (targetIndex !== -1) { tocItems.forEach(item => removeClass(item.element, 'active')); addClass(tocItems[targetIndex].element, 'active'); currentActive = targetIndex; } } // Re-check after browser might restore scroll position requestAnimationFrame(() => { requestAnimationFrame(() => { calculateRanges(); updateActiveSection(); }); }); // Event listeners window.addEventListener('scroll', handleScroll, { passive: true }); window.addEventListener('resize', handleResize); // Visual Viewport API for mobile viewport changes (keyboard, browser chrome) if (window.visualViewport) { window.visualViewport.addEventListener('resize', handleResize); window.visualViewport.addEventListener('scroll', handleResize); } // Browser back/forward navigation with hash changes window.addEventListener('popstate', () => { calculateRanges(); updateActiveSection(); }); // Handle dynamic content changes (if Webflow adds/removes content) const observer = new MutationObserver(debounce(() => { calculateRanges(); updateActiveSection(); }, 500)); observer.observe(tutorialContent, { childList: true, subtree: true, attributes: false }); // Recalculate after images load (affects layout positions) const images = tutorialContent.querySelectorAll('img'); images.forEach(img => { if (!img.complete) { img.addEventListener('load', handleResize, { once: true }); img.addEventListener('error', handleResize, { once: true }); } }); // Recalculate after fonts load if (document.fonts && document.fonts.ready) { document.fonts.ready.then(handleResize); } }); How to use Claude in Excel for HR: Headcount planningUse Claude to understand inherited spreadsheets, fix formula errors, and run what-if scenariosCategoryFinanceProductReading timeWatch time3min6minShareCopy linkhttps://claude.com/resources/tutorials/how-to-use-claude-in-excel-for-hr-headcount-planningClaude in Excel brings Claude directly into your spreadsheet workflow. It appears as a sidebar that can read your entire workbook—every tab, every formula, every relationship between cells—and help you understand, fix, and extend what's there.This tutorial walks through an HR headcount planning model: four tabs, formulas linking everywhere, and errors that need fixing. You'll see how to use plain language to get Claude to map a workbook you've inherited, trace and fix formula errors, run what-if scenarios, and build new calculations that reference data across sheets.What you can doUnderstand workbooks you didn't buildAsk Claude to explain how tabs connect. It reads everything first, then maps the data flow—which cells feed into which sheets, where assumptions live, how totals roll up.Find and fix formula errorsClaude catches broken references and traces where they cascade. Instead of hunting through tabs, you see every affected cell at once—and choose how to fix it.Run scenarios instantlyChange an assumption and see the impact across departments, totals, and downstream calculations. Claude recalculates and summarizes what shifted.Build formulas with business logicDescribe what you want calculated. Claude writes formulas that reference the right cells, apply conditional logic based on job levels or departments, and stay connected to your assumptions—not static values you'd have to update manually.Create visualizations from plain languageAsk for a pivot table or chart. Claude builds it from your data and tells you what stands out.In this videoFollow along with the video, or copy these prompts to try in your own workbook.0:51 — Get oriented in an inherited modelThe spreadsheet has four tabs—Assumptions, Headcount, Compensation, Summary—with formulas linking everywhere. Ask Claude to explain how it all connects before you start making changes."Walk me through this workbook—what's on each tab and how do they connect?"1:18 — Claude maps data flow and finds errorsClaude reads all four sheets, then returns a visual diagram showing how data flows from Assumptions through to Summary. It also catches #REF! errors you didn't ask about—and maps every cell they affect across three sheets.1:57 — Investigate a broken formulaA #REF! error is cascading through the model. Claude traces it to a deleted cell reference, analyzes the pattern in nearby cells, and figures out what the value should be."There's a #REF! error in cell E15 on the Headcount tab. What's causing it and can you fix it?"2:33 — Choose how to fix itClaude doesn't just fix the error—it explains the root cause, shows you the options (0, 1, or a custom value), and lets you decide. You stay in control of what changes.3:18 — Run a what-if scenarioChange one assumption and see the impact everywhere. Claude updates the model, then returns a before/after comparison by department so you can see which teams are hit hardest."Update the plan to assume 15% attrition instead of 10%. Show me what changes across departments."4:26 — Create a pivot table and chartDescribe the visualization you need. Claude builds a pivot table from your headcount data, adds a stacked bar chart, and surfaces insights—like which job level has the most people and where they're concentrated."Create a pivot table showing headcount by department and level, then add a stacked bar chart to visualize it."5:33 — Build a calculated column with business logicAsk for a new column and Claude writes actual formulas—not static values. Here, it pulls benefit rates from the Assumptions tab and applies different bonus percentages based on job level (10% for ICs, 15% for managers, 20% for directors)."Add a column that calculates the fully-loaded cost per employee using the rates on the Assumptions tab."What to noticeClaude reads before it answers. When you ask about the workbook, Claude doesn't guess—it reads all four sheets first, then explains how they connect. This prevents errors from incomplete context.Errors surface with their full impact. A single broken reference can cascade across multiple sheets. Claude shows you every affected cell so you understand the scope before you fix anything.Claude values your inputFor deeper understanding, ask Claude to explain its reasoning before making changes. You can always request that Claude walk through its thinking, and if something doesn't look right, undo (Cmd+Z or Ctrl+Z) reverses most actions.Formulas stay formulas. When Claude builds a new calculated column, it writes actual Excel formulas that reference your Assumptions tab and apply conditional logic. If your assumptions change later, the calculations update automatically.Get startedInstall Claude in Excel from the Microsoft AppSourceOpen it with Ctrl+Option+C (Mac) or Ctrl+Alt+C (Windows)Start with "Walk me through this workbook" on any spreadsheet you've inheritedClaude in Excel is available in beta on Max, Team, and Enterprise plans.Learn more:Claude in Excel help articleWhat is the Max plan?What you can doIn this videoWhat to noticeGet startedRelated tutorialsGetting started with Claude in ExcelGetting started with Claude in ExcelGetting started with Claude in ExcelTutorialTutorialTutorialHow to use Claude in Excel for accounting: Revenue model validationHow to use Claude in Excel for accounting: Revenue model validationHow to use Claude in Excel for accounting: Revenue model validationTutorialTutorialTutorialPrompting strategies for financial analysisPrompting strategies for financial analysisPrompting strategies for financial analysisTutorialTutorialTutorialClaude for financial services overviewClaude for financial services overviewClaude for financial services overviewTutorialTutorialTutorial @container (width < 52em) { .footer_layout { display: flex; column-gap: var(--_spacing---space--4rem); row-gap: var(--_spacing---space--4rem); } .footer_content_wrap.u-column-custom { display: contents; } } .footer_link { opacity: 1; transition: opacity 0.3s ease; } .footer_links_list:has(.footer_link:hover) .footer_link:not(:hover) { opacity: 0.4; } HomepageHomepageNextNextThank you! Your submission has been received!Oops! Something went wrong while submitting the form. (function () { const ROOT_SEL = "[data-prompt-scope]"; const TRIGGER_SEL = "[data-prompt-trigger]"; const MENU_ATTR = "data-prompt-menu"; const CLOSE_SEL = "[data-prompt-menu-close]"; const ACTION_SEL = "a, button, [data-prompt-menu-action]"; const ITEM_SEL = "[data-prompt-item], li, .menu-item, .w-dyn-item"; // ---------- Claude helper ---------- function buildClaudeUrl(text) { const url = new URL("https://claude.ai/new"); url.searchParams.set("q", text || ""); return url.toString(); } function openClaude(text) { const q = (text || "").trim(); if (!q) return; window.open(buildClaudeUrl(q), "_blank", "noopener"); } // ---------- Utilities ---------- let uid = 0; function makeId(prefix = "prompt-menu") { uid += 1; return `${prefix}-${Date.now().toString(36)}-${uid}`; } // Limit queries to elements that belong to THIS scope (ignore nested scopes) function qsaInScope(root, sel) { return Array.from(root.querySelectorAll(sel)).filter( (el) => el.closest(ROOT_SEL) === root ); } // Lightweight shield to stop click-through during the brief close animation function deployClickShield(ms = 300) { const sh = document.createElement("div"); sh.style.cssText = "position:fixed;inset:0;z-index:2147483647;pointer-events:auto;background:transparent"; document.body.appendChild(sh); setTimeout(() => { sh.remove(); }, ms); } // Pairing: find which menu a trigger controls (no manual ids needed) function resolveMenuForTrigger(root, trigger) { // 1) explicit data-prompt-trigger="x" -> [data-prompt-menu="x"] const explicit = trigger.getAttribute("data-prompt-trigger"); if (explicit) { const m = qsaInScope(root, `[${MENU_ATTR}="${explicit}"]`)[0]; if (m) return m; } // 2) aria-controls const ctrl = trigger.getAttribute("aria-controls"); if (ctrl) { const m = qsaInScope(root, `#${ctrl.replace(/(["'\\])/g, "\\$1")}`)[0]; if (m) return m; } // 3) nearest following sibling with [data-prompt-menu] let sib = trigger.nextElementSibling; while (sib && sib !== root) { if ( sib.hasAttribute && sib.hasAttribute(MENU_ATTR) && sib.closest(ROOT_SEL) === root ) return sib; sib = sib.nextElementSibling; } // 4) first menu in this root return qsaInScope(root, `[${MENU_ATTR}]`)[0] || null; } function ensurePairing(trigger, menu) { if (!menu.id) menu.id = makeId(); trigger.setAttribute("aria-controls", menu.id); if (trigger.tagName === "BUTTON" && !trigger.hasAttribute("type")) trigger.type = "button"; } // ---------- Animations (GSAP) ---------- function revealMenu(menu) { return new Promise((resolve) => { gsap.set(menu, { visibility: "visible", pointerEvents: "auto", willChange: "transform, opacity", }); gsap.killTweensOf(menu); gsap.fromTo( menu, { opacity: 0, scale: 0.96 }, { opacity: 1, scale: 1, duration: 0.28, ease: "power3.out", clearProps: "willChange", onComplete: resolve, } ); }); } function hideMenu(menu) { return new Promise((resolve) => { gsap.killTweensOf(menu); gsap.to(menu, { opacity: 0, duration: 0.2, ease: "power2.out", onComplete: () => { gsap.set(menu, { visibility: "hidden", pointerEvents: "none", clearProps: "opacity,scale,willChange", }); resolve(); }, }); }); } // ---------- Per-scope controller ---------- const stateMap = new WeakMap(); // root -> { openMenuEl, openTrigger, isAnimating } function getState(root) { let s = stateMap.get(root); if (!s) { s = { openMenuEl: null, openTrigger: null, isAnimating: false }; stateMap.set(root, s); } return s; } function setTriggersInteractive(root, enabled) { qsaInScope(root, TRIGGER_SEL).forEach((el) => { el.style.pointerEvents = enabled ? "auto" : "none"; }); } // ---------- Button fade animations ---------- function fadeOutButtons(root) { return new Promise((resolve) => { const buttons = qsaInScope(root, TRIGGER_SEL); if (buttons.length === 0) { resolve(); return; } gsap.killTweensOf(buttons); gsap.to(buttons, { autoAlpha: 0, duration: 0.2, ease: "power2.out", onComplete: resolve, }); }); } function fadeInButtons(root) { return new Promise((resolve) => { const buttons = qsaInScope(root, TRIGGER_SEL); if (buttons.length === 0) { resolve(); return; } gsap.killTweensOf(buttons); gsap.to(buttons, { autoAlpha: 1, duration: 0.2, ease: "power2.out", onComplete: resolve, }); }); } async function openMenuIn(root, menu, trigger) { const s = getState(root); if (s.isAnimating || s.openMenuEl === menu) return; s.isAnimating = true; // close any currently open in THIS scope only if (s.openMenuEl && s.openMenuEl !== menu) { await hideMenu(s.openMenuEl); if (s.openTrigger) s.openTrigger.setAttribute("aria-expanded", "false"); } // Fade out buttons when opening menu await fadeOutButtons(root); await revealMenu(menu); s.openMenuEl = menu; s.openTrigger = trigger || null; if (s.openTrigger) s.openTrigger.setAttribute("aria-expanded", "true"); // Disable only this scope's triggers while open setTriggersInteractive(root, false); s.isAnimating = false; } async function closeMenuIn(root) { const s = getState(root); if (!s.openMenuEl || s.isAnimating) return; s.isAnimating = true; deployClickShield(280); // prevent click-through during closing await hideMenu(s.openMenuEl); if (s.openTrigger) s.openTrigger.setAttribute("aria-expanded", "false"); s.openMenuEl = null; s.openTrigger = null; // Re-enable only this scope's triggers setTriggersInteractive(root, true); // Fade in buttons when closing menu await fadeInButtons(root); s.isAnimating = false; } // ---------- Initializer ---------- const initedRoots = new WeakSet(); function initScope(root) { if (!root || initedRoots.has(root)) return; initedRoots.add(root); if (!window.gsap) { console.warn("GSAP is required for prompt menus."); return; } // Normalize ALL menus in this scope to hidden on init (prevents “open on load” bugs) qsaInScope(root, `[${MENU_ATTR}]`).forEach((menu) => { if (menu.dataset.pmMenuInit) return; menu.dataset.pmMenuInit = "true"; gsap.set(menu, { visibility: "hidden", opacity: 0, scale: 1, pointerEvents: "none", }); // Optional close button const closer = menu.querySelector(CLOSE_SEL); if (closer) { if (closer.tagName === "BUTTON" && !closer.hasAttribute("type")) closer.type = "button"; closer.addEventListener("click", (e) => { e.preventDefault(); e.stopPropagation(); closeMenuIn(root); }); } // Menu item -> Claude menu.addEventListener("click", (e) => { const target = e.target.closest(ACTION_SEL); if (!target || !menu.contains(target)) return; const container = target.closest(ITEM_SEL) || target; const hiddenP = container.querySelector("[data-prompt-menu-text]"); if (!hiddenP) return; e.preventDefault(); e.stopPropagation(); const text = hiddenP.textContent || hiddenP.innerText || ""; openClaude(text); // close after action closeMenuIn(root); }); }); // Bind triggers (auto-wire) qsaInScope(root, TRIGGER_SEL).forEach((trigger) => { if (trigger.dataset.pmTrigInit) return; trigger.dataset.pmTrigInit = "true"; const menu = resolveMenuForTrigger(root, trigger); if (menu) ensurePairing(trigger, menu); trigger.setAttribute("aria-expanded", "false"); trigger.addEventListener("click", async (e) => { e.preventDefault(); const targetMenu = resolveMenuForTrigger(root, trigger); if (!targetMenu) return; const s = getState(root); if (s.openMenuEl === targetMenu) { await closeMenuIn(root); // toggle close return; } await openMenuIn(root, targetMenu, trigger); }); }); // Outside click (this scope only) document.addEventListener("click", (e) => { const s = getState(root); if (!s.openMenuEl) return; const insideMenu = s.openMenuEl.contains(e.target); const onTrigger = !!e.target.closest(TRIGGER_SEL) && root.contains(e.target); if (!insideMenu && !onTrigger) closeMenuIn(root); }); // ESC (this scope only) document.addEventListener("keydown", (e) => { if (e.key === "Escape") closeMenuIn(root); }); } // ---------- Boot + observe ---------- function boot(container = document) { const roots = container.querySelectorAll ? container.querySelectorAll(ROOT_SEL) : []; roots.forEach(initScope); } if (document.readyState === "loading") { document.addEventListener("DOMContentLoaded", () => boot(), { once: true, }); } else { boot(); } const mo = new MutationObserver((muts) => { for (const m of muts) { for (const node of m.addedNodes || []) { if (node.nodeType !== 1) continue; if (node.matches && node.matches(ROOT_SEL)) initScope(node); if (node.querySelectorAll) { node.querySelectorAll(ROOT_SEL).forEach(initScope); } } } }); mo.observe(document.documentElement, { childList: true, subtree: true }); })(); (function () { // ---------- Claude helper ---------- function buildClaudeUrl(text) { const url = new URL('https://claude.ai/new'); url.searchParams.set('q', text || ''); return url.toString(); } // ---------- Auto-grow helpers ---------- function sizeTextarea(el) { // Ensure UA defaults don't force a starting height el.setAttribute('rows', el.getAttribute('data-min-rows') || '1'); el.style.minHeight = '0px'; el.style.height = 'auto'; // allow shrink el.style.height = el.scrollHeight + 'px'; // fit content } function initAutogrow(root = document) { root.querySelectorAll('textarea[data-autogrow]').forEach((el) => { if (el.dataset.autogrowInit) return; el.dataset.autogrowInit = 'true'; // Respect your existing pattern for default text const preset = el.getAttribute('data-default-text'); if (preset != null && !el.value) el.value = preset; // Initial sizing sizeTextarea(el); // Resize as you type el.addEventListener('input', () => sizeTextarea(el)); // In case fonts/styles load late and change line-height window.addEventListener('load', () => sizeTextarea(el), { once: true }); }); } // ---------- Claude form initializer ---------- function initClaudeForms(root = document) { root.querySelectorAll('form[data-claude-form]').forEach((form) => { if (form.dataset.claudeInit) return; form.dataset.claudeInit = 'true'; const textarea = form.querySelector('[data-claude-textarea], textarea'); const trigger = form.querySelector('[data-claude-button], button[type="button"], a[data-claude-button]'); if (!textarea || !trigger) { console.warn('Claude form: missing textarea or trigger in', form); return; } // If the trigger is a