What do I get back from a ClearRun score?

Every score returns a verdict: a score from 0 to 100 (or no score if the text can't be scored), a grade, a risk level, a short summary, a verdict id and component signals. Paid plans receive four signals: Evidence support, Reasoning integrity, Confidence calibration and Ambiguity handling. Free use, developer API access and public lookups receive Evidence support and Reasoning integrity. Each successful hosted call also returns ClearRun's signature on the verdict and a receipt you can look up later. Mock scores have neither.

Last updated

The verdict

This is a real free-plan response from one score run on November 13, 2026. Scores can vary between runs.

{
  "verdict": {
    "verdictId": "6e07586d-e028-43d6-bec3-0f4821b22ca0",
    "schemaVersion": "1.2.0",
    "score": 52,
    "grade": "High risk",
    "risk": "elevated",
    "summary": "Review Evidence support.",
    "timestamp": "2026-11-14T01:02:12.552Z",
    "signature": "QsF676qMfyH0UUP2EkF7flPQ+JJuiJ254T2MbX5n0B8O6KCClYCSzFfecN/GNtVWtDqSD7cOzBNhpTvcN1AlDA==",
    "signatureKeyId": "sig-1",
    "signatureAlg": "ed25519",
    "signalsDigest": "21414c50f3bf2aa7033ebd80cc8aa297a5c2937a95bc30656ac543b2aad1737a",
    "signalsSalt": "fbf678c0c5cb43b2cc9ce4b199dc6191",
    "componentSignals": [
      {
        "name": "Evidence support",
        "level": 3,
        "note": "Limited supporting evidence"
      },
      {
        "name": "Reasoning integrity",
        "level": 6,
        "note": "Internally consistent"
      }
    ]
  },
  "receipt": {
    "receiptId": "rct_1d4b359fe4ca942d0bfa096b862a1031",
    "verdictId": "6e07586d-e028-43d6-bec3-0f4821b22ca0",
    "workspaceId": "account:6823bee4-6a1b-4864-b659-21f2eabd073a",
    "inputHash": "d2a0eedab8b7da41ed9bb2bf610cba76011ebad1d7108744c61b0838425511e9",
    "outputHash": "a7dd34a9d95011e53ac9d91baa9c37dd82cd19976e53e2a5c919696c9817075b",
    "verdictHash": "5137bdcc5be93c6b29074c5d4f93899118eeedd9b3aa90d47a22a1e2045789ee",
    "timestamp": "2026-11-14T01:02:12.598Z",
    "keyId": "sig-1",
    "alg": "Ed25519",
    "prevHash": "abe2c1a67ced960260f39d08606420a8d3b15ebd4c809456faef7590d4dd1071",
    "receiptHash": "255cf6a0965ca2956a02bcf96c16058038345e4de742070ddced78436c6204f8",
    "signature": "l1ipO7QSL6o90Iat5MR/Ti7yODCv6TaGrckXH3gEGDWOkfoKxnxFVFi6VDKGVi0RtqhjoHIuCCFHKcO45zHgDw==",
    "encKeyId": "enc-2026-10",
    "status": "stored"
  }
}
  • score: 0 to 100. Higher means the answer's confidence is better backed by its reasoning. It is not a truth score.
  • risk: low, moderate, elevated or high.
  • componentSignals: paid plans receive Evidence support, Reasoning integrity, Confidence calibration and Ambiguity handling. Free use, developer API access and public lookups receive Evidence support and Reasoning integrity.
  • summary: a short reason, in plain words.
  • grade: a short label for the score. Show it as returned.
  • verdictId: keep it. You use it to look the record up later.
  • signature, signatureKeyId, signatureAlg: ClearRun's Ed25519 signature on the verdict, and the key that goes with it.

The receipt

The receipt is ClearRun's record of this call. It holds fingerprints (hashes) of the text and the verdict, the time, and a link to the receipt before it.

Its status is usually stored. If it says queued, saving was delayed. You are not charged unless and until the receipt is saved. Once it is saved, the call is charged, because the score was made. The public lookup finds it once the save completes.

See What does signed and receipted mean?

Mock verdicts

Verdicts from mock mode (forceMock: true) are local test values, not real evaluations. Their signature fields are empty, there is no receipt, and the summary says the verdict is not a real evaluation.