Attestations
The signed daily Merkle roots, newest first, with everything a check needs.
GET /api/v1/attestationscurl https://vitna.ai/api/v1/attestationsResponse
{
"attestations": [ { "coversOn": "2026-08-22", "root": "...", "statement": "...", "signature": "...", "publicKey": "...", "leafCount": 918, "signedAt": "..." } ],
"algorithm": "...",
"statementPrefix": "...",
"howToVerify": [
"Rebuild each observation's leaf as sha256(0x00 || canonical JSON) with the field order in platform/witness/attest.mjs.",
"Hash pairs as sha256(0x01 || left || right), promoting an odd node unchanged, until one root remains.",
"Check that root against this row, then verify the signature over `statement` with `publicKey`, Ed25519."
],
"proves": "The record for that day has not changed since it was signed.",
"doesNotProve": "That the provider published the price. ...",
"meta": { "generatedAt": "...", "source": "database", "count": 3 }
}proves and doesNotProve ship in the payload
Both sentences are in the response rather than only in documentation, because this is the claim most easily overstated by someone quoting the record.
A signature proves the record for that day has not changed since it was signed. It does not prove the provider published the price: no provider signs its own pricing page. What ties a figure to a provider is the archived document, which records what Vitna fetched.
Getting from one figure to a root
This endpoint publishes roots. `/observations/{id}/proof` publishes the path from a single observation up to one, which is what turns a root from a claim into something checkable.
Serving the snapshot
Answers 200 with attestations: null, the algorithm and statement prefix still present, and a note pointing at the live record.
null rather than [], for the same reason as everywhere else: an empty list would read as "nothing was ever signed".
Parameters
None.
Caching
public, max-age=60, stale-while-revalidate=3600.
The endpoint is published in full because a root nobody can read attests to nothing. The mechanism is that the statement was in the open on the day, so an edit made later has to contradict something already published.
