Skip to content

ygwyg/agent-experience

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Agent Experience

A practical reference to the patterns, surfaces, and design principles behind AI agents — from primitives to production.

https://agent-experience.dev

What is this?

Agent Experience is an open-source reference site covering 26 patterns across five categories:

  • Foundations — Mental models like tool use, ReAct, planning, and reflection
  • Patterns — Concrete architectures like MCP, computer use, multi-agent orchestration, and skills
  • Infrastructure — The plumbing: memory, context management, sandboxes, guardrails, and harnesses
  • Surfaces — Where agents meet humans: generative UI, IDE agents, CLI agents, and chat interfaces
  • Design — Making agents usable: human-in-the-loop, observability, and evaluation

Each pattern includes a description, key ideas, real-world examples, and links to go deeper.

Running locally

bun install
bun dev

Building

bun run build

Deploying

The site deploys to Cloudflare Workers with static assets:

bun run build
wrangler deploy

Tech stack

  • React 19 + Vite
  • Motion (Framer Motion) for animations
  • Plain CSS
  • Cloudflare Workers (static assets) for hosting

Contributing

Contributions are welcome! A few ground rules:

  • This is not an advertising platform. PRs that exist primarily to promote a company or product will not be merged. We want to showcase the best, most useful examples — not the ones with the biggest marketing budget.
  • PRs are not guaranteed to be merged. We review everything and prioritize quality and relevance. Don't take it personally if we pass.
  • Focus on what's genuinely best-in-class. If you're adding an example, it should be something practitioners actually use and recommend. Include a URL and a concise, honest description.
  • Keep descriptions factual. No superlatives, no marketing copy. Say what it does, not how great it is.

Good PR examples

  • Adding a new example that's widely adopted and genuinely useful
  • Fixing outdated information (dead links, incorrect descriptions)
  • Adding a missing pattern that belongs in the reference
  • Improving the site's accessibility or performance

PRs we'll close

  • Adding your company's product as an example when it's not widely known or adopted
  • Marketing-style descriptions ("the world's leading...", "the most powerful...")
  • Removing competitors' examples to make room for yours

License

MIT

About

A visual reference to the patterns, surfaces, and infrastructure behind AI agents.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors