For Agents
Public, read-only surfaces for AI agents and tools.
How to read
- Start at
llms.txtfor the site map and when-to-use notes. - Prefer
Accept: text/markdown, or a.mdpath (for example/about.md). - Auth is none; detail in
auth.md. - Machine surfaces are listed in
openapi.jsonand the RFC 9727 API catalog.
Surfaces
| Resource | Purpose |
|---|---|
llms.txt |
Site map and when-to-use notes. |
Accept: text/markdown, .md |
Markdown alternate for HTML pages. |
openapi.json |
OpenAPI description of the public paths. |
/.well-known/api-catalog |
RFC 9727 linkset. |
auth.md |
Authentication notes (none required). |
/.well-known/api-status.json |
Service status. |
/.well-known/ard.json,
ai-catalog.json
|
Agentic Resource Discovery catalog. |
/.well-known/agent-skills/index.json
|
Published skills (limit / cursor pagination). |
sitemap.xml |
Full URL list for crawlers. |
Authentication
None. No accounts, API keys, tokens, OAuth, or hosted MCP server. Everything in
llms.txt
and
sitemap.xml
is public over HTTPS.
Rate limits
No published quota. Keep rates polite;
robots.txt
asks for Crawl-delay: 2.
Capabilities
-
WebMCP in-page tools (project search, page markdown) where the browser supports it.
The
local-relay embed is off by default (loopback or
?webmcp=1). No hosted MCP endpoint. -
Agent skills at
/.well-known/agent-skills/index.json.
Aliases
/developers, /docs, /api,
/getting-started, and /setup redirect here.