For custom stacks
Custom and self-hosted agent readiness
Rails, Django, Laravel, Astro, Hugo, a hand-rolled Node server, or anything else you control end to end: if you can add a file to the document root and a route to the server, every one of the 14 checks is reachable. The gap is never the framework, it is that agent readiness was not on anyone's list until now.
Native ceiling
Best achievable score without fronting
No Cloudflare fronting needed
Native ceiling is already 100
What agents currently see
This varies more than any other category, since it covers everything from a mature Rails monolith to a static site generator with three pages. The common pattern: robots.txt and sitemap.xml usually exist in some form (frameworks and static-site generators generate them by convention), SSL is typically already correct, but agents.md, llms.txt, an MCP server card, and /.well-known/verified.md are almost never present, because they are new conventions no framework ships by default yet.
Structured data and content negotiation are the two checks that split hardest by maturity: a team that already cares about SEO usually has Organization or WebSite JSON-LD in place; a team that does not usually has neither that nor any handling for a non-HTML Accept header.
An MCP server card is the check that correlates most with whether a site already exposes an API. If you already ship a public or partner API, adding a card at /.well-known/mcp.json describing it as tools is often a half-day job on top of documentation you already maintain. If you do not have an API, this check simply is not relevant yet, and the scanner treats a clear absence the same as a missing file: something to add when it makes sense, not a hard requirement.
Your ceiling, and why
100. Full server and filesystem access means there is no platform-level wall for any of the 14 checks, only implementation work: static files for agents.md, llms.txt, and llms-full.txt, a route or middleware for content negotiation and the MCP card, a rewrite rule or reverse-proxy config for AI-crawler-friendly robots.txt.
Which delivery model fits
The Agent Ready Kit is usually the fastest path here: since you already have full repo and server access, paste-ready instructions (exact file paths, a route handler snippet, the .well-known entries) are something your existing team can merge in an afternoon. Done for you fits if the team's bandwidth is the actual blocker, not the access.
What is typically failing
- agents.md
- llms.txt / llms-full.txt
- MCP server card
- .well-known/verified.md
- Content negotiation (Accept: text/markdown)
- x402 payment signaling
See where you stand
Run a free scan, or start with the self-serve kit if you already know you need it.