AGENTS.md — Cross-Platform Agent Instructions

# AGENTS.md — Cross-Platform Agent Instructions

> This file provides instructions for AI agents that read AGENTS.md (GitHub Copilot, Cursor, Windsurf, Cline, Aider, OpenCode, and others).

## About This Repository

This is a **prompts and agents library** with 480+ prompts and 180+ specialized agents for AI development. Organized in 28 thematic categories.

## How to Use

### Finding Prompts

Prompts are organized in `library-prompts/categories/` by topic number:

| # | Category | Best For |
|---|----------|----------|
| 01-08 | Core Development | Planning, SQL, testing, docs, .NET, Java, Python, Azure |
| 09-15 | Agent & Workflows | Agents, Git/CI, refactoring, MCP, Power Platform, utils, blueprints |
| 16-19 | Community Prompts | OpenAI examples, Claude library, Reddit community |
| 20-24 | Specialized | Niches, prompt engineering, image gen, role-play, advanced |
| 25 | AI Skills & Agents 🆕 | SKILL.md creation, multi-agent, MCP servers, OpenSkills |
| 26 | ML/AI Models 🆕 | GPT-5.2 migration, Claude Opus 4.6, HuggingFace, LoRA |
| 27 | DevOps 🆕 | Docker Compose, GitHub Actions, Railway, IaC, security |
| 28 | Ecosystem 🆕 | SkillsMP, MCP Market, GitHub Topics, tool comparison |

### Using Agents

Agents are in `library-agents/` as `.agent.md` files. Each defines a specialized AI persona.

Key agents:
- `agent-selector.agent.md` — Choose the right agent for any task
- `skill-architect.agent.md` — Create professional SKILL.md files
- `multi-agent-orchestrator.agent.md` — Coordinate multiple agents
- `codebase-analyst.agent.md` — Codebase health analysis
- `security-guardian.agent.md` — Security auditing
- `ml-pipeline-engineer.agent.md` — ML/AI pipeline management
- `devops-deployer.agent.md` — Deployment and infrastructure
- `ecosystem-scout.agent.md` — Discover new AI tools and skills
- `release-manager.agent.md` — Release management

## Related Repository

- [garri333/Skills](https://github.com/garri333/Skills) — 245+ reusable skills for AI agents

## Conventions

- Prompts use YAML frontmatter: title, version, category, tags, author, description, language
- Use Conventional Commits: `feat:`, `fix:`, `docs:`, `refactor:`
- Agents are defined in `.agent.md` files with specialized instructions