//scanner
Is your site ready for AI agents?
Measures, layer by layer, whether a site is readable, citable and operable by AI agents: robots.txt, llms.txt, JSON-LD, manifest, MCP server and consent.
What to fix first
The six layers
//L0
Access
robots.txt: are answer bots allowed? Training is a separate decision.
//L1
Context
llms.txt, llms-full.txt and agents.md, with no model-directed instructions.
//L2
Fact
JSON-LD on the home page: entity, offer, prices consistent across layers.
//L3
Capability
Manifest: /.well-known/ucp or /.well-known/mcp.json.
//L4
Operation
An MCP server that answers tools/list.
//L5
Consent
Action tools with declared human approval.
The scanner uses its own identity and never impersonates other bots: each bot's access is read from robots.txt. Firewall blocks are not tested. Action tools are never called on third-party sites. Results are cached for 30 minutes per domain; limit of 10 scans per hour.