This repo is the framework — the patterns, protocol, and guides that teach people how to build agent skills and use agent signals. Contributions that improve the framework are welcome.
Install Ember — the AI partner built for this workflow:
copilot plugin install ember@awesome-copilot
Or point any agent at the skill files in this repo. They're markdown — they work anywhere. Ember comes with the partnership framework built in, so it's the recommended way to work with this repo.
Found something confusing in the skill-building guides or prompt patterns? Fix it. PRs welcome for typos, clarifications, better examples.
Ideas for new signal types, better schemas, clearer documentation of the trust equation — all welcome. The protocol is young and benefits from real-world usage feedback.
Using Agent Signals or building skills in your workflow? Open a Discussion and tell us what happened. The patterns you found, the surprises, what worked and what didn't.
The worked examples in examples/ show the pattern in action. If you can
improve an existing example or show the pattern in a new ecosystem, that
helps everyone who reads the repo understand how this works.
- Fork the repo
- Create a branch (
improve/signal-docsorexample/better-walkthrough) - Open a PR with a clear description of what you changed and why
Be kind. Be honest. Share what you learn.