Thank you for your interest in contributing to Awesome Agentic Patterns! This repository aims to be the most comprehensive, production-grade guide for AI agents, MCP tooling, and developer workflows.
You can contribute in several ways:
- Adding an Architectural Pattern: Submit a diagram and real-world failure/success analysis of an agent pattern (e.g., Critic-Actor, MapReduce agents, Memory-Augmented loops).
- Adding an MCP Server: Introduce a battle-tested MCP server with connection parameters and configuration snippets.
- Adding Rules & Prompt Templates: Share verified
.cursorrules,CLAUDE.md, orAGENTS.mdtemplates tailored for specific frameworks (e.g., Flutter, Rust, PyTorch). - Improving the Setup CLI: Extend
scripts/setup_agent.pyto auto-detect more project frameworks.
When submitting an MCP server to the directory:
- Must have an active repository or npm/PyPI package.
- Must include a concise 1-sentence description of permissions/safety implications.
- Must provide the exact CLI launch command (
uvxornpx).
When submitting .cursorrules or CLAUDE.md:
- Focus on actionable constraints rather than generic praise ("Write clean code").
- Include instructions on how the agent should handle edge cases, testing, and file integrity.
- Keep token size reasonable (under 500 tokens per rule file).
- Fork the repo and create your branch from
main:git checkout -b feature/my-new-agent-pattern
- Verify linting and syntax:
python scripts/setup_agent.py --help
- Commit your changes using Conventional Commits:
feat: add Anthropic Swarm handoff patterndocs: update MCP Postgres server entryfix: correct typo in CLAUDE.md template
- Push to your fork and submit a Pull Request to
DYNOSuprovo/awesome-agentic-patterns. - Ensure all CI checks pass.
Thank you for making the AI engineering ecosystem better!