Awesome Agentic Patterns is a curated catalogue of production-ready AI agent design patterns.
The project addresses the gap between toy tutorials and real-world agent deployments by collecting repeatable patterns that help autonomous and semi-autonomous AI agents accomplish useful work. Each pattern must be repeatable across multiple teams, agent-centric in improving how agents sense reason or act, and traceable to a public reference such as a blog post talk repository or paper. The catalogue organizes patterns into eight categories including Context & Memory, Feedback Loops, Learning & Adaptation, Orchestration & Control, Reliability & Eval, Security & Safety, Tool Use & Environment, and UX & Collaboration.
Developers building production agents should use this resource to discover and compare proven approaches rather than inventing solutions from scratch. The project suits teams shipping autonomous systems who need concrete reference implementations and architectural guidance. The accompanying website at agentic-patterns.com provides discovery tools beyond the README including a pattern explorer with filtering and search, a side-by-side comparison tool, an interactive decision guide, graph visualization of pattern relationships, curated pattern packs for common agent architectures, and developer guides on pattern selection.
The project maintains an actively curated collection with patterns stored in a structured folder system that auto-generates the README tables, indicating ongoing maintenance and organization. The codebase is built with Astro and deployed on Vercel with source code in the apps/web directory, showing a modern web-first approach to making patterns discoverable. The README explicitly invites community contributions through pull requests and notes that categories remain fluid, suggesting the maintainers view the taxonomy as evolving with community input.
