Machine-readable docs for AI agents
Anyknown publishes machine-readable versions of all documentation pages so that Local Builder agents (Claude Code, Codex, Gemini CLI) can load the full SDK surface into context without a browser.
Available endpoints
Section titled “Available endpoints”| Endpoint | Contents |
|---|---|
/llms.txt | Index of available pages with titles and descriptions — the starting point for agents |
/llms-small.txt | Curated corpus: every generally-available page, with coming-soon pages excluded |
/llms-full.txt | Complete corpus: full prose of every page, including coming-soon pages |
/<page-slug>.md | Raw markdown for a single page, e.g. /capabilities/object-storage.md |
Pages with coming-soon status are excluded from the /llms.txt index and from /llms-small.txt because they do not describe working functionality yet. They do appear in /llms-full.txt (the complete corpus), where each one opens with an explicit coming-soon notice in its body prose — treat those pages as a contract preview, not a working feature.
Fetch the index and pipe it to your agent:
curl https://docs.anyknown.com/llms.txtFetch a specific capability reference:
curl https://docs.anyknown.com/capabilities/object-storage.mdFetch the full docs for broader context:
curl https://docs.anyknown.com/llms-full.txtMCP server
Section titled “MCP server”If you use the @anyknown/mcp MCP server with Claude Code, the server has direct access to your App’s source tree.
For platform documentation, use the curl approach above to load relevant capability pages into your agent’s context before building.