Agent QA · Docs + read-only runtime · Free

Can an agent understand and safely reach your API?

Score an OpenAPI document, compare it with a baseline, or run bounded checks against public HTTPS endpoints. Runtime tests accept only GET and HEAD, never accept credentials, and return redacted evidence.

Add to GitHub ActionsAPI reference

Current OpenAPI

JSON only, up to 1 MB. No URL is fetched.

Readiness report

Score, gate conditions, and regression details.

Load the sample or paste an OpenAPI document to begin.

Optional regression baseline

Paste a prior Agent QA JSON report or earlier OpenAPI document. Leave blank for a one-off score.

Safe runtime verification

Enter 1–5 JSON test cases. Public HTTPS only; GET/HEAD only; no headers, credentials, or request bodies. Redirects are revalidated and responses are bounded.

Download runtime workflow
Runtime results appear here.

Regression signal

Compare score changes plus new and resolved findings against a saved report or specification.

Scheduled monitoring

Founding Pro can keep ten checks running; Teams shares up to 100 with roles, history, and failure alerts. Manage monitoring →

Agent access

Use REST /api/agent-qa/runtime or MCP verify_api_runtime_readonly.

curl --fail -X POST https://agentmediatools.com/api/agent-qa/compare \
  -H 'Content-Type: application/json' \
  --data-binary @agent-qa-request.json > agent-qa-result.json
node -e "const r=require('./agent-qa-result.json'); process.exit(r.gate.passed ? 0 : 1)"

Download GitHub Actions workflow