Skip to main content

Crovia — The Global Standard for AI Training Data Attribution. Cryptographic proofs, absence verification, and automatic royalty settlement.

Project description

Crovia Core Engine (Open Core)

If AI is trained on your data, there should be a receipt.

Crovia is an offline-verifiable evidence engine for declared AI training data.

It does not accuse.
It does not infer intent.
It does not enforce compliance.

Crovia produces deterministic artifacts that anyone can verify independently.


Read this first (30 seconds)

If you read only one thing, read this:

Crovia turns a declaration into a closed, verifiable evidence capsule.
Nothing more. Nothing less.

No trust required.
No network required.
No hidden logic.


What Crovia produces (CRC-1)

Crovia generates a CRC-1 Evidence Pack — a closed set of files that fully describe:

  • what was declared
  • what was produced
  • how integrity can be verified

Each CRC-1 pack contains:

  • receipts.ndjson — declared training receipts
  • validate_report.md — deterministic validation result
  • hashchain.txt — integrity hash-chain
  • trust_bundle.json — normalized trust summary
  • MANIFEST.json — authoritative artifact contract

All files are offline-verifiable.

Specification:
docs/CROVIA_ARTIFACT_SPEC.md


Try it (single command)

Generate a complete CRC-1 Evidence Pack:

# Unified CLI
crovia run --period 2025-11 --budget 100000

# Or with custom receipts
crovia run \
  --receipts examples/minimal_royalty_receipts.ndjson \
  --period 2025-11 \
  --out out_crc1

Other commands:

crovia status    # Check system status
crovia test      # Run module tests
crovia score     # Calculate Crovia Score

This creates a fully self-contained evidence capsule in out_crc1/.

  • No network
  • No secrets
  • Fully deterministic

Inspect the artifacts

Example:

ls out_crc1
cat out_crc1/MANIFEST.json

MANIFEST.json defines exactly which files must exist.

Nothing implicit.
Nothing hidden.


Verify evidence (offline, by anyone)

Verification requires only the files themselves.

Example:

crovia-verify out_crc1

Expected result:

✔ All artifacts present
✔ trust_bundle JSON valid
✔ Hashchain verified

✔ CRC-1 VERIFIED

If verification fails, the evidence is invalid.

No trust assumptions.
No authority required.


Design principles

  • Offline-first
  • Deterministic
  • No attribution claims
  • No enforcement logic
  • Evidence > opinions

Crovia produces facts, not judgments.


Where to see real evidence

Crovia Open Core does not ship conclusions.

All public, inspectable evidence generated with this engine lives here:

https://github.com/croviatrust/crovia-evidence-lab

That repository contains:

  • reproducible CRC-1 capsules
  • offline-verifiable artifacts
  • neutral semantic observations (DSSE)
  • presence / absence observations (Spider)

If you want to see results, go there.
If you want to reproduce them, stay here.


License

Apache-2.0
CroviaTrust

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

crovia-0.2.0.tar.gz (26.9 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

crovia-0.2.0-py3-none-any.whl (29.1 kB view details)

Uploaded Python 3

File details

Details for the file crovia-0.2.0.tar.gz.

File metadata

  • Download URL: crovia-0.2.0.tar.gz
  • Upload date:
  • Size: 26.9 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.3

File hashes

Hashes for crovia-0.2.0.tar.gz
Algorithm Hash digest
SHA256 f000ed19f0e1f2926e67b9cc95f1fc78bfccf66474eccb5f331e72e41d044a04
MD5 3b3a1d8e1217388be22332871a642481
BLAKE2b-256 8c65bb642ef110d6fc36e758e83348fa59d7c1a1b7bbf4f2490d910cca4cbdf6

See more details on using hashes here.

File details

Details for the file crovia-0.2.0-py3-none-any.whl.

File metadata

  • Download URL: crovia-0.2.0-py3-none-any.whl
  • Upload date:
  • Size: 29.1 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.3

File hashes

Hashes for crovia-0.2.0-py3-none-any.whl
Algorithm Hash digest
SHA256 83b05d9863328901a59758fd6de203762ddd4c38a796af40dcc7fa7213447a30
MD5 3ce4a14d7423fe1e4d1690c5242490f4
BLAKE2b-256 03b14d58de1d5927222759de1c26b4526ad7b92bc19b1613534ed92716dd2507

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page