Rule catalog
Readiness rule catalog
Every check behind Technical, Estimated and Verified Agent Readiness, grouped by category, with what each rule detects and how to fix it.
Discoverability
lowexperimentalinfoexperimentalinfoexperimentalmediumexperimentalinfoexperimentalmediumexperimentalinforequiredmediumrequiredinfoexperimentalinfoexperimentalcriticalrequiredhighrequiredcriticalrequiredmediumexperimentallowexperimentallowexperimentalinforequiredlowexperimentallowexperimentalinfoexperimentalinforequiredinforequiredlowexperimentallowexperimentallowrequiredlowrequired
Legitimate verifiable agent traffic is blocked by the observed edge policy
agent-edge-001
Anonymous automation is challenged; verified agents pass; spoofed claims are refused
agent-edge-003
A cryptographically verifiable agent request reached the site
agent-id-001
A claimed agent identity was trusted without verification
agent-id-002
Signature headers reached the origin intact
agent-id-003
Identity verification is not fail-closed
agent-id-005
Declared AI-crawler policy
ai-bot-policy-001
Declared allow but agent request blocked
ai-bot-policy-003
Anonymous and declared-agent access differ
ai-bot-policy-005
AI-training usage group conflicts with another declaration
ai-bot-policy-006
AI crawlers blocked in robots.txt
ai-crawler-blocked
Page crawlability blocked
blocked-crawlability
Bot protection blocking AI agents
bot-protection-blocking-agents
Content negotiation response inconsistent with its Content-Type
content-neg-002
Markdown negotiation works but omits Vary: Accept
content-neg-004
Content-Signal directives are malformed
content-signal-001
No Content-Signal directives in robots.txt
content-signals-absent
Orphaned pages within the scanned sample
crawl-graph-001
Pages buried deep within the scanned sample
crawl-graph-002
Dead-end pages within the scanned sample
crawl-graph-005
No agent-relevant Link headers
link-headers-absent
llms.txt missing or invalid
llms-txt-missing-or-invalid
Consolidated representation preference applied but response is unchanged
machine-rep-004
Consolidated representation preference applied but Vary omits Prefer
machine-rep-005
No Markdown representation for agents
markdown-for-agents-absent
Sitemap missing or invalid
sitemap-missing-or-invalid
Structured data
lowrequiredcriticalrequiredhighrequiredhighrequiredlowrequiredcriticalrequiredinforequiredmediumrequired
Poor heading hierarchy
heading-hierarchy-poor
Availability missing from Offer
missing-availability
Missing Offer data in Product schema
missing-offer-data
Missing Product schema
missing-product-schema
Page token cost excessive for agents
page-token-bloat
Price missing or invalid in Offer
price-missing-invalid
No semantic HTML5 landmarks
semantic-html-absent
Variant information missing from structured data
variant-info-missing
Consistency
Actionability
mediumexperimentalmediumexperimentalmediumexperimentalinforequiredhighrequiredmediumexperimentallowexperimentalmediumrequiredmediumrequiredlowexperimentalmediumexperimentalinfoexperimentalinfoexperimentallowexperimentallowexperimentalmediumexperimentallowexperimental
MCP tool metadata contains prompt-injection language
agent-meta-001
MCP tool name collision or reserved-token shadowing
agent-meta-004
MCP tool metadata contains hidden or obfuscated characters
agent-meta-008
No agent interface discovery surface
agent-interface-discovery-absent
JS-only critical content
js-only-critical-content
MCP endpoint requires auth but has no discoverable OAuth metadata
mcp-auth-001
MCP endpoint answered initialize without an MCP-Protocol-Version response header
mcp-obs-001
MCP discovery declares a dead endpoint
mcp-discovery-invalid
MCP endpoint absent
mcp-endpoint-absent
MCP discovery endpoint completed the handshake but failed tool enumeration
server-card-004
MCP server identity declared vs. runtime mismatch
server-card-005
MCP tools include write/destructive/external-side-effect risk classes
tool-risk-001
Site exposes an in-page WebMCP surface
webmcp-001
WebMCP surface present but its tools could not be enumerated
webmcp-002
WebMCP tool missing schema or description
webmcp-003
WebMCP tool metadata contains prompt-injection or obfuscated language
webmcp-004
WebMCP tool safety annotations contradict themselves
webmcp-005
Trust
mediumexperimentalhighexperimentalhighexperimentalmediumrequiredmediumrequiredmediumrequired
Referenced domain does not resolve
llms-supply-002
Instructions pipe a remote script into a shell
llms-supply-006
Install instructions name a claimable package or domain
llms-supply-008
Missing OpenGraph meta for citation
missing-og-meta-for-citation
Return policy unavailable
missing-returns-policy
Shipping information unavailable
shipping-info-unavailable
UCP
inforequiredlowrequiredmediumrequiredmediumrequiredmediumrequired
No agent payment protocol beyond UCP
agent-payments-absent
Universal Commerce Protocol: Checkout capability missing
ucp-checkout-missing
Universal Commerce Protocol profile absent
ucp-profile-absent
Universal Commerce Protocol profile invalid
ucp-profile-invalid
Universal Commerce Protocol: signing keys invalid
ucp-signing-keys-invalid
Citability
Scovant Core, the open-source static scanner, implements 57 of these 66 rules; its CLI gathers evidence for 48 today (9 of them only when the site publishes no MCP file). The other 9 need Scovant Cloud's agent execution. Its 69 checks are diagnostics, mapped to these rules in the crosswalk. More on /open-source.