Inspect first.
Integrate with a contract.
Start with a record, a public test key and the checks a recipient can run independently of the service.
One field. Two ways to read it.
Select a field to connect its meaning to the exact bundled fixture.
Read the evidence offered for the claim; its presence is not a causal execution proof.
[{
"claim": "This demonstration record carries source and signal information",
"evidence_text": "This demonstration record carries source and signal information for inspection.",
"claim_support": 1.0
}]This is the same synthetic fixture used in the Record Lab. Download original JSON · Reference tools
Run the same checks outside the service.
The fixture is synthetic, English and signed with a public test key. It contains no production inference.
python3 verify.py sample/receipt.json --pubkey sample/pubkey.txt
node verify.ts sample/receipt.json --pubkey sample/pubkey.txtPython supports machine-readable output using --json; the Node verifier does not implement that output mode. See the bundled README for requirements.
Integrity is one check—not the final interpretation.
- 01
Source and signal disclosure: inspect claims, evidence and method identifiers.
- 02
Canonical record integrity: recompute the declared hash under its canonicalisation rules.
- 03
Signature: verify the signed hash string against the provided test key.
- 04
Interpretation: preserve nulls and distinguish measured values from heuristic or unavailable signals.
Test-key note. A bundled public test key establishes only the demonstration’s cryptographic consistency. Production use requires a trusted key-distribution and lifecycle process.
Agree the deployed contract for the selected environment.
Endpoint, authentication, schema, disclosure tier, transport responsibilities and acceptance tests must be explicit. Missing values must not become zeros.