create-azure-agent · diff
git:20260321.b056f9f to git:20260328.04e9e1c
22 added, 42 removed. Audit A to A.
---
name: create-azure-agent
- description: >
- Interactive initialization script that generates Azure AI Foundry Agent API deployment
- wrappers. Trigger with "deploy this skill to azure", "create an azure foundry agent",
- "scaffold bicep and python for azure", or when the user wants to take a local skill
- and deploy it as a hosted service in Azure AI Foundry.
+ description: Deploy a skill as an Azure AI Foundry hosted agent
+ argument-hint: "[skill-dir]"
allowed-tools: Bash, Write, Read
---
- ## Dependencies
-
- This skill requires **Python 3.8+** and standard library only. No external packages needed.
-
- **To install this skill's dependencies:**
- ```bash
- pip-compile ./requirements.in
- pip install -r ./requirements.txt
- ```
-
- See `../../requirements.txt` for the dependency lockfile (currently empty — standard library only).
-
- ---
- # Azure AI Foundry Agent Scaffolder
-
- You are an expert Cloud Integration Architect. Your job is to convert local Agent Skills into deployable Azure AI Foundry Agent Services.
-
- Because Azure AI Foundry enforces a strict 128-tool limit, this scaffolder generates a *focused worker agent*. The generated python service (`azure_agent.py`) will precisely parse the target `SKILL.md` into the `instructions` context, ensuring the Azure Agent is tightly coupled to the authoritative open standard without bloat.
+ Follow the `create-azure-agent` skill workflow to generate Azure AI Foundry deployment
+ wrappers for an existing agent skill.
- ## Execution Flow
+ ## Inputs
- Execute these phases in order. Do not skip phases.
+ - `$ARGUMENTS` — optional path to the skill directory to deploy. Omit to start with discovery.
- ### Phase 1: Guided Discovery
- Ask the user for the parameters for the Azure scaffolding:
- 1. **Target Skill**: The directory path to the existing skill (e.g., `plugins/my-plugin/skills/my-skill`).
- 2. **Naming Preference**: Should the Azure Project and Cosmos DB instances use a specific prefix, or generate automatically?
+ ## Steps
- Wait for the user's answers before generating any files.
+ 1. If `$ARGUMENTS` provides a skill directory, resolve and validate the path
+ 2. Follow the create-azure-agent phased workflow: confirm the target skill, gather Azure
+ configuration (subscription, resource group, region, naming preferences), then run
+ `scaffold_azure_agent.py` to generate Bicep templates and the Python Azure AI Projects
+ SDK deployment wrapper
+ 3. Summarize generated files in the skill's `azure_deployment/` directory
+ 4. Instruct on reviewing `.bicep` parameters and running `az deployment group create`
- ### Phase 2: Action Scaffold
- Once approved, use bash execution to run the scaffold script:
+ ## Output
- ```bash
- # Example invocation
- python ${CLAUDE_PLUGIN_ROOT}/scripts/scaffold_azure_agent.py \
- --skill [target-skill-path]
- ```
+ `azure_deployment/azure_agent.py` (Azure AI Projects SDK orchestration script) and
+ `azure_deployment/main.bicep` (Cosmos DB, AI Search, and Foundry Project infrastructure).
- ### Phase 3: Post-Scaffold Review
- After successful execution, summarize the outputs generated within the target skill's `azure_deployment/` directory:
- 1. `scaffold_azure_agent.py` - The `azure-ai-projects` Python SDK orchestration script.
- 2. `main.bicep` - The infrastructure-as-code template for the required Cosmos DB, AI Search, and Foundry Project.
+ ## Edge Cases
- Advise the user to review the `.bicep` parameters and run `az deployment group create` when they are ready to provision the infrastructure. Offer to run `audit-plugin` to validate the underlying skill before they deploy.
+ - If `$ARGUMENTS` is empty: ask for the target skill directory before proceeding
+ - If Azure credentials are not configured: instruct user to run `az login` first
+ - Azure AI Foundry enforces a 128-tool limit — scaffold generates a focused worker agent
+ - Offer to run `/agent-scaffolders:audit-plugin` to validate the skill before deploying