An education and information site about Agentic Resource Discovery (ARD) — the open, web-native way for any AI agent, on any platform, to discover and verify the capabilities it needs: other agents, MCP servers (tools), and skills.
Built with MkDocs and Material for MkDocs.
- Site: https://search4agents.github.io/
- Catalog (spec well-known root): https://search4agents.github.io/.well-known/ai-catalog.json
- Fun skills:
pip install -r requirements.txt
mkdocs serve # live preview at http://127.0.0.1:8000
mkdocs build --strict # production build into ./siteThe two fun skills are real application/agentskill+zip artifacts. Edit the
source in skills/<name>/SKILL.md, then repackage the zips:
python3 skills/build_skills.py- Manual:
mkdocs gh-deploy --cleanbuilds the site and pushes it to thegh-pagesbranch. - Automated: pushing to
mainruns.github/workflows/docs.yml, which builds (--strict) and deploys to GitHub Pages. - Scheduled: the same workflow also rebuilds and redeploys every morning at
9:00 a.m. America/Los_Angeles. GitHub cron is UTC and ignores daylight
saving, so it triggers at 16:00 and 17:00 UTC and a
gatejob runs the build only when it is actually 9am Pacific.
Pick a single GitHub Pages source to match your method: Deploy from a branch →
gh-pagesfor the manual route, or GitHub Actions for the automated one.
Search 4 Agents is a federated registry. To list your own agentic services,
publish an ai-catalog.json on your domain and open a pull request adding a row
to docs/registry.md. The full step-by-step plan and
acceptance checklist live in docs/register.md.
Licensed under the Apache License 2.0. Feature requests are welcome — please open a Pull Request.