{"service":"agent-action-risk","version":"1.0.0","status":"ok","one_liner":"Score the runtime risk of any agent action (shell / SQL / HTTP) before it executes.","why_this_exists":"Anthropic publishes the \"Execute actions with care\" principle. Every agent builder needs a production gate. This API is that gate, in 30ms, with no LLM call.","endpoints":{"POST /score":"Evaluate one action. Returns risk_score, decision, reversibility, blast_radius, required_approvers, reasons.","POST /batch":"Evaluate up to 50 actions at once.","GET /policy":"List the active policy thresholds and rule families.","GET /openapi.json":"OpenAPI 3.1 schema.","GET /healthz":"Liveness.","GET /quota":"Free tier limits."},"pricing":{"free":"1000 calls/day per IP","pro":"$9/mo unlimited + custom rules (mailto:coindany4@gmail.com)","team":"$29/mo per seat — policy authoring + audit log + SSO"},"example_curl":"curl -X POST https://api.lazy-mac.com/agent-action-risk/score -H \"content-type: application/json\" -d '{\"kind\":\"shell\",\"command\":\"rm -rf /\"}'"}