llms-full.txt@packages/website/public · git:20260826.3df3c2b · 2026-08-26 · sha256 46618c42bd11dab9
llms-full.txt@packages/website/public git:20260826.3df3c2bA

Name: llms-full.txt@packages/website/public
Author: smith-horn
Immutable. This exact content is served forever at /api/v1/blob/46618c42bd11dab9.
# Skillsmith - Complete Reference for AI Systems
> Extended documentation for LLM-powered answer engines and AI assistants.
> For the summary version, see: https://www.skillsmith.app/llms.txt

## Product Overview

Agent builders need the same vetted skills, not one-off files nobody's reviewed. Skillsmith shares them across your team, scans them for risk before they ship, and shows you which ones are actually being used. It also provides a searchable registry of hundreds of thousands of indexed skills, reusable prompts, workflows, and configurations that extend AI assistant capabilities.

Skills are YAML files installed to `~/.claude/skills/` that provide:
- Domain-specific knowledge and best practices
- Reusable workflows and templates
- Tool configurations and integrations
- Project-type-specific guidance

## Installation

### MCP Server (Recommended)

Add to your MCP client configuration (e.g., `~/.claude/settings.json`):

```json
{
  "mcpServers": {
    "skillsmith": {
      "command": "npx",
      "args": ["-y", "@skillsmith/mcp-server"]
    }
  }
}
```

Restart your MCP client after adding the configuration.

### CLI Alternative

```bash
npm install -g @skillsmith/cli
skillsmith search testing
skillsmith install jest-helper
```

## MCP Tools (API)

When using the MCP server, these tools are available:

| Tool | Parameters | Description |
|------|-----------|-------------|
| `search` | query, category, trust_tier, min_score, limit | Search skills by keyword or semantic query |
| `get_skill` | id (format: author/name) | Get detailed skill information |
| `install_skill` | id | Install skill to ~/.claude/skills/ |
| `uninstall_skill` | id | Remove an installed skill |
| `recommend` | context (optional) | Get project-based skill recommendations |
| `validate` | path | Validate a skill's YAML structure |
| `compare` | ids (2-5 skill IDs) | Compare skills side-by-side |
| `skill_suggest` | project_path, current_file, recent_commands, error_message, installed_skills, limit, session_id | Proactively suggest skills based on current project context (counts against monthly quota) |
| `skill_outdated` | include_deps | Check installed skills for staleness and dependency satisfaction status |
| `index_local` | force, skillsDir | Index skills from ~/.claude/skills/ or a custom directory |
| `skill_publish` | skill_path, check_references, reference_patterns, create_repo, visibility, add_topic | Prepare a local skill for publishing |
| `skill_rescan` | skillName (optional) | Re-scan an installed skill's content for security/quality changes |
| `skill_recover_source` | homeDir, only, embedding | Recover the canonical GitHub source of locally-installed skills (read-only) |
| `inventory_push` | (none) | Push this machine's installed-skill inventory to your Skillsmith account for the web dashboard |
| `skill_updates` (Individual+) | skillIds (optional) | Check the registry for newer skill versions |
| `skill_diff` (Individual+) | skillId, oldContent, newContent, oldRiskScore, newRiskScore, hasLocalModifications, trustTier | Section-level diff between two versions of a skill |
| `skill_pack_audit` (Individual+) | pack_path, check_trigger_quality | Audit all skills in a directory |
| `skill_audit` (Team+) | skillIds (optional) | Check skills for known security advisories |
| `skill_inventory_audit` (Team+) | deep, homeDir, projectDir, applyExclusions | Audit every installed AI coding client's skill inventory for local namespace collisions |
| `apply_namespace_rename` (Team+) | auditId, collisionId, action (apply/custom/skip/revert), customName, confirmed | Apply a rename suggestion from an inventory audit |
| `apply_recommended_edit` (Team+) | auditId, collisionId, confirmed | Apply a recommended prose edit from an inventory audit |
| `undo_apply` (Team+) | count, suggestion_id | Session-scoped undo of the most recent apply_namespace_rename/apply_recommended_edit changeset(s) |
| `team_workspace` (Team+) | action (create/list/get/delete), name, description, workspaceId | Manage team workspaces |
| `share_skill` (Team+) | action (add/remove/list), workspaceId, skillId | Add, remove, or list skills in a team workspace |
| `publish_private` (Team+) | skillId | Mark a skill private on this device, hiding it from your own search results |
| `team_analytics_dashboard` (Team+) | period | Per-user tool usage counts, top tools, daily trend |
| `team_usage_report` (Team+) | period, format | Weekly/monthly usage summary with period comparison |
| `audit_export` (Enterprise) | startDate, endDate, eventType, limit | Export audit log events for a time range |
| `audit_query` (Enterprise) | actor, resource, eventType, result, startDate, endDate, limit | Query audit logs with filters |
| `siem_export` (Enterprise) | startDate, endDate, format | Export audit events for SIEM ingestion |
| `analytics_dashboard` (Enterprise) | period, includeRecommendations | Recommendation accuracy, adoption curves, team aggregation |
| `usage_report` (Enterprise) | period, format | Comprehensive usage report with all metrics |
| `configure_sso` (Enterprise) | action (set/test/remove), idpMetadataUrl, idpEntityId, protocol | Configure SSO/SAML integration |
| `sso_settings` (Enterprise) | includeMetadata | View current SSO/SAML configuration |
| `private_registry_publish` (Enterprise) | skillId, version, content | Publish a skill version to your team's private registry as a pending submission requiring admin approval |
| `private_registry_manage` (Enterprise) | action (list/get/install/deprecate/undeprecate/namespace/submissions/approve/reject), skillId | Manage your team's private registry, including reviewing pending submissions |
| `rbac_manage` (Enterprise) | action (create_role/list_roles/get_role/delete_role), name, roleId, permissions, description | Manage RBAC roles |
| `rbac_assign_role` (Enterprise) | action (assign/revoke/list_assignments), userId, roleId | Assign or revoke roles for users |
| `rbac_create_policy` (Enterprise) | action (create/list/get/delete), name, policyId, effect, resources, actions | Create and manage RBAC access policies |
| `compliance_report` (Enterprise) | format (soc2/cyclonedx/json), period, includeUserActivity, backfillDependencies | Generate SOC2, CycloneDX SBOM, or JSON compliance reports |

## CLI Commands

| Command | Description | Example |
|---------|-------------|---------|
| `skillsmith search <query>` | Search for skills | `skillsmith search testing` |
| `skillsmith install <id>` | Install a skill | `skillsmith install jest-helper` |
| `skillsmith uninstall <id>` | Remove a skill | `skillsmith uninstall jest-helper` |
| `skillsmith list` | List installed skills | `skillsmith list` |
| `skillsmith recommend` | Get project recommendations | `skillsmith recommend` |
| `skillsmith info <id>` | Show skill details | `skillsmith info jest-helper` |
| `skillsmith compare <ids>` | Compare skills | `skillsmith compare jest-helper vitest-helper` |
| `skillsmith sync` | Sync skill registry | `skillsmith sync` |
| `skillsmith status` | Show sync status | `skillsmith status` |
| `skillsmith config` | Manage configuration | `skillsmith config set apiKey sk_live_...` |

Short alias: `sklx` (e.g., `sklx search testing`)

## Authentication

### Methods (in priority order)

1. **Personal API Key** (recommended): `X-API-Key: sk_live_*` header or config file
2. **Supabase Anon Key**: Built-in fallback, 30 requests/minute
3. **No Auth**: 10 trial requests total

### API Key Configuration

Option 1 — Config file (cross-platform):
```json
// ~/.skillsmith/config.json
{ "apiKey": "sk_live_your_key_here" }
```

Option 2 — MCP server environment:
```json
{
  "mcpServers": {
    "skillsmith": {
      "command": "npx",
      "args": ["-y", "@skillsmith/mcp-server"],
      "env": { "SKILLSMITH_API_KEY": "sk_live_your_key_here" }
    }
  }
}
```

Note: Shell `export` does NOT work for MCP servers. They run as subprocesses and don't inherit shell environment.

## Pricing

| Tier | Price | API Calls/Month | Rate Limit |
|------|-------|-----------------|------------|
| Community | Free | 100 | 30/min |
| Individual | $9.99/month | 1,000 | 60/min |
| Team | $25/user/month | 10,000 | 120/min |
| Enterprise | Custom (Contact Sales) | Unlimited | 300/min |

Annual billing saves 17% (pay for 10 months, get 12).

Free 14-day trial available for Individual and Team plans.

## Trust Tiers

| Tier | Description | Badge |
|------|-------------|-------|
| verified | Official reviewed skills from trusted authors | Green |
| community | Community-reviewed and tested skills | Blue |
| experimental | New or beta skills, less vetting | Yellow |

## Skill Schema Format

Skills are YAML files with this structure:

```yaml
---
name: my-skill
description: What the skill does
version: 1.0.0
author: username
category: development
tags: [testing, react]
tools: [Read, Write, Edit, Bash]
---

# Skill content (Markdown)
Instructions and prompts for the AI assistant...
```

## Frequently Asked Questions

Q: What are agent skills?
A: Skills are reusable prompts and workflows installed to ~/.claude/skills/ that give AI assistants specialized capabilities like testing, deployment, code review, and documentation generation.

Q: Is Skillsmith free?
A: Yes. The Community tier is free forever with 100 API calls per month. Paid plans start at $9.99/month for developers who need more.

Q: What AI tools does it work with?
A: Skillsmith works with any MCP-compatible client including Claude Code, Claude Desktop, Cursor, Windsurf, and other tools that support the Model Context Protocol.

Q: How do I install a skill?
A: Add the Skillsmith MCP server to your settings.json, then ask your assistant "Install the [skill-name] skill." Skills install to ~/.claude/skills/ in one step.

Q: What counts as an API call?
A: Each skill search, recommendation request, or skill installation counts as one API call. Viewing cached results does not count.

Q: Can I change plans at any time?
A: Yes, upgrade or downgrade at any time. Changes take effect at the next billing cycle.

Q: What happens if I exceed my API limit?
A: You'll be notified at 80% and 90% usage. Once exceeded, calls are rate-limited until the next cycle or until you upgrade.

Q: Can I use Skillsmith offline?
A: Installed skills work offline. Searching, recommendations, and updates require an internet connection.

Q: How do I report bugs or request features?
A: Open an issue at github.com/smith-horn/skillsmith/issues. Enterprise customers have dedicated support channels.

Q: Where are skills installed?
A: Skills install to ~/.claude/skills/ by default and are automatically available in all sessions.

Q: What is the private registry, and how does publishing get approved?
A: The private registry (Enterprise tier) is a shared, org-scoped skill store — publish a skill once and every teammate can find and install it. A published version sits as a pending submission, invisible to search, get, and install (even to the person who published it) until a different team admin approves or rejects it. Self-approval is blocked at the database level. Both approve and reject are terminal — a rejected version can only be superseded by publishing a new version.

## Packages

| Package | Install | Description |
|---------|---------|-------------|
| @skillsmith/mcp-server | `npx -y @skillsmith/mcp-server` | MCP server for AI assistants |
| @skillsmith/cli | `npm i -g @skillsmith/cli` | Command-line interface |
| @skillsmith/core | Internal | Core library (not published separately) |

## Blog

Engineering and product articles covering Skillsmith development and AI-native workflows.

- [Building Skillsmith with Claude Flow V3](https://www.skillsmith.app/blog/building-skillsmith-claude-flow): How multi-agent orchestration — SPARC methodology, hive-mind execution, wave-based delivery, and Plan-Review — shipped 159 commits at 94% goal achievement across 534 Claude Code sessions.
- [From GitHub to Search Results: How Skillsmith Indexes Skills](https://www.skillsmith.app/blog/how-skillsmith-indexes-skills): The five-stage pipeline that discovers, validates, scans, scores, and indexes skills for semantic search.
- [Composing Agents, Sub-Agents, Skills, and Sub-Skills](https://www.skillsmith.app/blog/agent-skill-framework): A decision framework for when to use each component and how to compose them without context overflow.
- [Security and Trust: How Skillsmith Scans and Quarantines Skills](https://www.skillsmith.app/blog/security-quarantine-safe-installation): How static analysis, trust tiers, and quarantine protect users from malicious skills.

## Links

- Website: https://www.skillsmith.app
- Documentation: https://www.skillsmith.app/docs
- Private Registry: https://www.skillsmith.app/docs/private-registry
- Governance: https://www.skillsmith.app/docs/governance
- Skills Browser: https://www.skillsmith.app/skills
- Pricing: https://www.skillsmith.app/pricing
- GitHub: https://github.com/smith-horn/skillsmith
- Contact: https://www.skillsmith.app/contact
- Terms: https://www.skillsmith.app/terms
- Privacy: https://www.skillsmith.app/privacy

## License

Elastic License 2.0 (ELv2)
- Self-hosting: Permitted
- Modification: Permitted for own use
- Managed service: Not permitted (cannot offer Skillsmith as a service to third parties)