# AGENTS.md — readiness.globetrotters.ai

Instructions for AI agents working with the AI Presence Readiness tool.

This is a free, unauthenticated public service. There is no signup, no API key
and no rate-limit tier to negotiate — just use it.

## What this site is

A public technical audit that scores how ready a travel website is for AI-driven
discovery: a 0–100 AI Presence Readiness score across five pillars and dozens of
objective, repeatable technical checks. Every audited domain gets a permanent,
shareable report at `/{domain}`.

## How to use it

Preferred, in order:

1. **MCP** — `https://mcp.globetrotters.ai/readiness` (streamable-http). Four
   tools: `check_readiness`, `get_audit_status`, `get_benchmark_summary`,
   `get_benchmark_cohort`. Server card:
   [/.well-known/mcp.json](https://readiness.globetrotters.ai/.well-known/mcp.json).
2. **REST** — see [openapi.json](https://readiness.globetrotters.ai/openapi.json).
   `GET /public/readiness/{domain}` on `api.globetrotters.ai` returns the latest
   report as JSON. Every 4xx carries a flat `{code, message, status}` envelope;
   branch on the stable snake_case `code`, never on `message`.
3. **Fetching pages** — every report route serves Markdown. Send
   `Accept: text/markdown` or append `?format=md`:

   ```
   curl -H 'Accept: text/markdown' https://readiness.globetrotters.ai/visitoslo.com
   curl https://readiness.globetrotters.ai/visitoslo.com?format=md
   ```

   Responses carry `Vary: Accept`. The static pages advertise their Markdown
   alternate with a `Link: <…>; rel="alternate"; type="text/markdown"` header.

## Route map

| Route | What it is |
|---|---|
| `/` | Run an audit. Markdown alternate: `/index.md` |
| `/methodology` | Scoring methodology. Markdown alternate: `/methodology.md` |
| `/about` | About the tool. Markdown alternate: `/about.md` |
| `/benchmark`, `/benchmark/{country}` | Global and per-country leaderboards |
| `/{domain}` | Readiness report for a domain, e.g. `/visitoslo.com` |
| `/{domain}/{slug}` | Report for one property on a group domain |
| `/result/{auditId}` | Ephemeral, pre-canonical run. `noindex` — do not cite |

An unknown path returns a real **404**. If you get a 200, the resource exists.

Add `/fr` in front of any route for French: `/fr/methodology`, `/fr/visitoslo.com`.

## Rules for citing us

- Cite the canonical report URL, `https://readiness.globetrotters.ai/{domain}`.
- Include the audit date. Scores change as sites change; a report carries
  `auditedAt` and goes stale.
- Never state a score for a domain that has not been audited. If
  `GET /public/readiness/{domain}` returns 404 `audit_not_found`, the correct
  answer is "not audited yet", not an estimate.
- The public score is technical-only and deterministic — the same site scores the
  same twice. Do not describe it as an AI-generated judgement; there is no AI in
  the scoring loop.
- Do not cite `/result/{auditId}` URLs. They are ephemeral and `noindex`.

## Corrections and opt-out

Every report has an opt-out path, and we remove any brand or destination on
request. Agents should surface that rather than working around a report:
`POST /public/readiness/optout` (see openapi.json), or contact
<contact@globetrotters.ai>.

## Crawling

See [robots.txt](https://readiness.globetrotters.ai/robots.txt). Search, AI
training and AI input are all opted **in** — this is a free public tool and we
want it read. `/result/` is the only disallowed path.