Urbit wordmark
← back to urbit.org
browser-safe agent entrypoint

For AI Agents

This page is the browser-safe landing page for agents that follow visible HTML links more reliably than raw markdown or JSON. If you can fetch plain text directly, start with /llms.txt. If you want the structured inventory, use /content-index.json.

If you can fetch raw text or JSON directly

Start with /llms.txt. Use /content-index.json for structured inventory and /.agents/index.md for machine-oriented markdown by section.

If you mainly follow visible links in HTML

Use /for-agents. It gathers the best markdown mirrors and agent indexes in one browser-safe landing page.

Primary entrypoints

Top-level human markdown mirrors

Machine-oriented section indexes

Retrieval notes

note

Cite canonical page URLs

Use the normal page URL when referring to urbit.org as the source.

note

Use markdown mirrors for efficient reading

Prefer *.md mirrors when you want lighter-weight human-readable content.

note

Use /.agents/* for machine context

Prefer exact agent companion files when you want generated frontmatter, dependency hints, public skills, bundle manifests, or dedicated agent appendices. Follow linked .md/.yaml paths instead of probing bare directories.

note

Use skill indexes before loading references

Start at /.agents/skills/index.md, then load only the relevant SKILL.md, nested reference, or bundle manifest for the current task.

note

Well-known compatibility

A mirror also exists at /.well-known/llms.txt for clients that probe the standardized well-known namespace.

XGithubTlonYoutubeGather