Metadata-Version: 2.4
Name: inferencebench-envelope
Version: 0.0.2
Summary: Canonical signed-envelope spec + Sigstore signer/verifier for InferenceBench
Project-URL: Homepage, https://github.com/yobitelcomm/bench
Project-URL: Documentation, https://yobitelcomm.github.io/bench/concepts/envelope
Author-email: Yobitel Communications <bench@yobitel.com>
License: Apache-2.0
Keywords: ai,benchmark,ml,reproducibility,sigstore
Classifier: Development Status :: 2 - Pre-Alpha
Classifier: Intended Audience :: Developers
Classifier: Intended Audience :: Science/Research
Classifier: License :: OSI Approved :: Apache Software License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.12
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Classifier: Topic :: Security :: Cryptography
Requires-Python: >=3.12
Requires-Dist: cryptography~=44.0
Requires-Dist: pydantic~=2.9
Requires-Dist: sigstore~=3.5
Requires-Dist: uuid7~=0.1
Description-Content-Type: text/markdown

# inferencebench-envelope

The canonical signed-envelope spec for InferenceBench results. This is the product's defensibility moat — every benchmark result is signed via Sigstore and verifiable by anyone with `bench verify`.

## Status

Phase 1 active development. Schema v1 in progress (ticket 0004).

## Concepts

- **Envelope**: Pydantic v2 model representing one benchmark run's full provenance
- **Hardware fingerprint**: SHA-256 of DMI UUID + GPU PCI IDs + serials + driver + BIOS state
- **Software provenance**: pip freeze hash + git commit + (optional) container image digest
- **Signature**: keyless OIDC via Sigstore cosign + Rekor transparency log entry

See [docs/concepts/envelope.md](../docs/concepts/envelope.md) for the full conceptual guide.
