☰ Documentation menu
Introduction
What is Provenova? Getting started
Concepts & workflows
Core concepts Capturing runs Reproduce & drift Result cards & badges Compliance & attestations Corpus & leaderboard
Using the app
Product tour (every page & button) Pricing FAQ
Libraries & reference
Libraries & downloads CLI reference API reference Frameworks reference Open schemas (qlprov) Deployment & self-hosting Licensing
Libraries & reference

Open schemas (qlprov)

Provenova's provenance format is an open, versioned standard called qlprov, not a proprietary blob. The schemas are JSON Schema (draft 2020-12) and ship inside provenova-core at provenova_core/schemas/qlprov/. Because the format is open and self-verifying, a run you export can be read, validated, and hash-checked by anyone — no Provenova server required.

qlprov/run/1.0

The full record of a run: its circuit, backend identity, calibration snapshot, compilation metrics, result distribution, and the hashes that bind them (run_hash, chain hashes). This is what ql show <id> renders and what GET /api/v1/runs/<id> returns.

Self-verifying. An exported qlprov/run/1.0 document recomputes its own run_hash offline:

from provenova_core import verify_run_hash
assert verify_run_hash(document)   # True iff the leaf hashes still produce run_hash

The run_hash is the SHA-256 of a Merkle root over the seven labeled leaf hashes embedded in merkle.leaves (schema, circuit, compilation, backend, calibration, execution, results) — any alteration to those bound hashes changes the recomputed root and the check fails. The human-readable body fields can be validated by cross-checking them against the leaves: circuit.content_sha256, backend.identity_hash, and calibration.content_sha256 equal their leaves directly, and the results leaf is the Merkle root of each result's counts_sha256. The hash is also stable across the local and hosted stores, which is why it doubles as the idempotency key for ql push.

qlprov/calibration/1.0

The normalized calibration-snapshot format: per-qubit T1 / T2 / readout error, per-gate error rates, and capture metadata. Every vendor's native calibration is normalized to this shape on capture and by the crawler, which is what makes cross-vendor comparison (the leaderboard) possible.

Why open schemas matter

  • Portability — your provenance isn't locked into one vendor or one server.
  • Verifiability — trust the math, not the platform: recompute the hash yourself.
  • Interoperability — tools, auditors, and the public corpus all speak the same format.

Next: Deployment & self-hosting.