# Churchill Coffee auth.md

This document describes how automated agents may access **Churchill Coffee** (`https://churchill.coffee`).

## Audience

Agents answering questions about Churchill Coffee (a coffee shop in Brisbane, Australia) or discovering the public read-only JSON on this origin.

## Agent access

Churchill Coffee does **not** operate an OAuth authorization server, MCP server, or self-service agent credential issuer. There is no `POST /agent/auth` registration endpoint.

Public pages and the JSON listed in [`/openapi.json`](https://churchill.coffee/openapi.json) are **anonymous**. Send HTTP GET requests directly.

<a id="anonymous-public-api"></a>

### Anonymous access

Use for:

- HTML or markdown homepage (`GET /`, prefer `Accept: text/markdown`)
- Site facts (`GET /api/site.json`)
- Health (`GET /api/health.json`)
- Discovery files (`/robots.txt`, `/sitemap.xml`, `/llms.txt`, `/.well-known/api-catalog`)

## Methods

| Method             | How                                                                       |
| ------------------ | ------------------------------------------------------------------------- |
| Anonymous HTTP GET | No `Authorization` header. Tokens are not issued and are ignored if sent. |

## Related discovery

- API catalog: [`/.well-known/api-catalog`](https://churchill.coffee/.well-known/api-catalog)
- ARD manifest: [`/.well-known/ai-catalog.json`](https://churchill.coffee/.well-known/ai-catalog.json)
- Human notes: [`/docs/api`](https://churchill.coffee/docs/api)