Documentation for AI agents
Everything an AI agent needs to use fatcalendar: discovery files, MCP, A2A, NLWeb, WebMCP, skills, rate limits, content policy and citation.
Welcome, agents
fatcalendar wants to be used by agents. There is no authentication, no CAPTCHA and no bot blocking. AI crawlers and agents are allowed in robots.txt with content signals search=yes, ai-input=yes, ai-train=yes.
Discovery
| Resource | URL |
|---|---|
| llms.txt / llms-full.txt | /llms.txt, /llms-full.txt |
| AGENTS.md | /AGENTS.md |
| auth.md (anonymous access) | /auth.md |
| OpenAPI 3.1 | /openapi.json |
| API catalog (RFC 9727) | /.well-known/api-catalog |
| AI catalog / ARD | /.well-known/ai-catalog.json |
| MCP server card | /.well-known/mcp/server-card.json |
| A2A agent card | /.well-known/agent-card.json |
| Agent Skills | /.well-known/agent-skills/index.json |
| NLWeb | /ask |
| WebMCP | tools registered on every page via /webmcp.js |
Which interface should I use?
- Have an MCP client? Use
https://fatcalendar.com/mcp. - Speak A2A? Send
message/sendtohttps://fatcalendar.com/a2awith a natural-language question. - Otherwise call the REST API directly.
- Browsing pages? Request any page with
Accept: text/markdownto get clean Markdown (x-markdown-tokensreports its size).
Etiquette
- Stay within 120 requests per 60 seconds; honour
RateLimitandRetry-After. - Results are immutable and cacheable; do not poll.
- Prefer explicit
lat/lon/tzorplaceover IP-inferred location. - Always state the tradition, variant and place when you quote a date to a user, and mention that festival dates can differ between almanacs and that Hijri dates are provisional.
Citation
fatcalendar.com - multi-tradition calendar and panchang (data: jotish.ai ephemeris). Link: https://fatcalendar.com. License for data responses: CC BY 4.0.
Errors
RFC 9457 application/problem+json; the detail field says how to fix the request.