Agent-Ready Site Audit

# Agent-Ready Site Audit

## Purpose
Teach an outside bot how to audit a public site for discoverability by MCP clients and other bots: well-known files, clear link relations, and honest machine-readable entry points. Adapted for the Dead Brands Skills Hub from the Apache-2.0 agent-ready-cloudflare skill in https://github.com/fabricioctelles/skills (skill agent-ready-cloudflare), which builds on public Cloudflare agent-ready guidance.

## When to use
- Before advertising that bots can join a site.
- When llms.txt, MCP URLs, or .well-known files may be missing or stale.
- After a domain or path migration.

## Limits
- Report only what you can fetch. No invented scores or fake grade seals.
- Do not claim affiliation with Cloudflare or any scanner brand.
- Security: read-only audit unless the owner asked for a fix.

## Steps
1. Fetch `/`, `/llms.txt`, `/robots.txt`, `/sitemap.xml`, and any documented MCP URL.
2. Check for `.well-known` manifests the site claims to publish. Record 200/404 honestly.
3. Verify link relations and canonical URLs on key pages.
4. Confirm machine-readable lists (skills, tools, games) match the human pages.
5. Note broken redirects, mixed content, and pages that block bots unintentionally.
6. Produce a prioritized fix list with exact paths. Re-fetch after fixes (3 passes).

## Examples
Input: audit https://mcp.example.com
Output:
```
llms.txt 200 — lists MCP + skills count
/sitemap.xml 200
/.well-known/agent.json 404 — add or stop advertising
/hub/skills count matches D1 published
```

## Credit
Adapted under Apache-2.0 from https://github.com/fabricioctelles/skills (agent-ready-cloudflare). Hub packaging by Dead Brands. License retained in import sources.