This repository provides a comprehensive catalog of "skills" – code snippets, templates, and configurations – designed to enhance AI coding agents, particularly those working with Azure SDKs and Microsoft AI Foundry. It aims to ground these agents with domain-specific knowledge, enabling them to generate more accurate and relevant code for developers building on Microsoft platforms. The primary benefit is increased AI agent productivity and precision in specialized development contexts.

How It Works

The project organizes over 124 skills into language-specific directories (Python, .NET, TypeScript, Java), each with a distinct suffix for automatic discovery. These skills are intended to be selectively integrated into an agent's environment, either via direct installation (npx skills add), manual cloning, or through platforms like Context7. The repository also includes MCP (Microsoft Cognitive Platform) server configurations for LLM tool integration and AGENTS.md templates to define role-specific agent personas and behaviors. The core approach leverages these structured skills as activation context to surface relevant domain knowledge within large language models.

Quick Start & Requirements

• Primary Install: Use npx skills add microsoft/agent-skills (requires Node.js and npm/npx).
• Alternative Installation: Manual git clone and copy/symlink skills into your project's .github/skills/ directory.
• Context7 Installation: npx ctx7 skills install /microsoft/agent-skills azure-ai-projects-py (requires Context7 CLI).
• Prerequisites: Node.js is required for npx commands. Specific skills may have underlying Azure service dependencies.
• Links:
  • Context7 Catalog: context7.com/microsoft/agent-skills
  • Generated Docs: microsoft.github.io/agent-skills

Highlighted Details

• A catalog of 124 skills across Python, .NET, TypeScript, and Java, covering Azure AI, data, messaging, security, and more.
• Pre-defined custom agents for roles like backend, frontend, infrastructure, and planner.
• MCP server configurations for documentation, development tools (GitHub, Playwright), and utilities.
• Daily updated "live documentation" indexed by Context7 for semantic search.
• A robust testing framework with 1144 scenarios and acceptance criteria for all skills.
• Iterative code improvement via the "Ralph Loop" and Sensei-style scoring.
• A project scaffolder for full-stack Azure AI Foundry applications using Vite, React, FastAPI, and Azure Container Apps.

Maintenance & Community

The repository features automated GitHub Actions workflows for daily documentation updates and PR generation, indicating active maintenance. Specific community channels (like Discord/Slack) or core contributor details are not explicitly detailed in the provided README.

Licensing & Compatibility

• License: MIT.
• Compatibility: The MIT license permits broad use, including commercial applications and linking within closed-source projects, with standard attribution requirements.

Limitations & Caveats

Users are strongly advised to install skills selectively to prevent "context rot," which can dilute agent attention, increase token usage, and conflate patterns. The effectiveness of these skills is dependent on the AI agent's capability to interpret and apply the provided context effectively.