Auth-protected — its tools can't be enumerated without credentials. Category: Other.
Spec conformance20%50
✗ initialize returned no protocolVersion
✓ initialize HTTP 401
Security (OWASP MCP)30%75
auth-protected — tool surface not enumerable unauthenticated (connect via proxy for behavioral grading)
Reliability / performance20%92
single-probe latency 403ms (PRELIMINARY — continuous uptime/p95 from proxy telemetry lands in v2)
Tool hygiene15%50
tool surface not enumerable unauthenticated
Transparency / provenance15%80
HTTPS ✓
RFC 9728 oauth-protected-resource metadata ✓
Observed behavior
No proxied traffic observed for this host yet. Connect it at /connect and its grade gains a measured Reliability score + per-tool behavioral evidence — the half a static scan can't produce.
Findings
INFOMCP02 OAuth-protected (401 + auth challenge). Behavioral grading (annotation-truthing, real latency) requires connecting via the wmcp.sh proxy — v2.
We re-grade mcp.cirra.ai on a schedule and alert your Slack/webhook the moment its tools change or its grade drops — rug-pull insurance for the connection.
Share this report card
A 1200×630 card with the grade + audit — drop it in a post, Slack, or your repo.
A live badge — it re-verifies itself and shows current stability. Static scorecards can't. Paste it in your README or site to show users you're independently audited.
Add the wmcp.sh trust oracle as an MCP server and call grade_mcp_server / check_mcp_drift in your agent's pre-connection gate:
https://wmcp.sh/mcp/trust
How this grade is computed. An open, independent rubric — Spec conformance (20%), Security mapped to the OWASP MCP Top 10 (30%), Reliability (20%), Tool hygiene (15%), Transparency (15%) — run by connecting to the server and inspecting its real MCP surface. The grade is free and identical whether or not the operator pays. v1 uses static + spec signals from a single connection; continuous uptime, real latency, and annotation-truthing (declared readOnly vs observed behavior) layer on via the wmcp.sh proxy.