score.lastseen.dev

Derived trust scores for MCP servers, built on the lastseen.dev register — every score carries its depth, its resolution confidence, and a methodology reference. A separate service from the register: the register publishes dated observations and sells no score; this service derives a trust_score on top of that public record (separation of functions).

Read the score with its depth

Depth is part of the number, not metadata: a score is shown as e.g. 80 · depth 2/5 · provisional. A depth-2 80 means “alive, everything else unknown” — mortality, not merit. See /methodology.

Endpoints (no key, GET; MCP over POST)

curl https://score.lastseen.dev/v1/trust_score/mcp/io.github.06ketan/medium-ops
curl https://score.lastseen.dev/v1/schema_history/mcp/io.github.06ketan/medium-ops
curl https://score.lastseen.dev/v1/dependency_risk/mcp/io.github.06ketan/medium-ops
# MCP (streamable HTTP, JSON-RPC over POST):
curl -X POST https://score.lastseen.dev/mcp -H 'Content-Type: application/json' -H 'Accept: application/json, text/event-stream' -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"trust_score","arguments":{"id":"io.github.06ketan/medium-ops"}}}'

The three FACT/VALUE surfaces: trust_score (the only scored surface, carries a methodology ref in every response), schema_history and dependency_risk (notary-class dated facts, no scoring).

Machine surfaces

Everything this service serves is listed, not left to be guessed — /openapi.json (schema) · /docs (reference) · /llms.txt (index for agents) · /sitemap.xml (human pages) · /healthz (liveness) · /prices.json — the free-tier limits and the floor promise this service publishes; its charging_status is NONE: a declaration of terms, not a request for payment.

See also the State of the Agent Economy report, published on the register.