code-ownership-audit

When you pull in an open-source snippet and tweak a few lines, legally it may still be bound by its upstream license. AI-assisted coding makes this ambiguity far more common—you may not even realise how closely a generated implementation mirrors existing code. code-ownership-audit compares your code against a reference at every AST node, reports the longest shared expression, and tags each risk with an exemption rationale: a common idiom, a shape dictated by interface conventions, or a genuine copy of expression. The entire engine relies only on the Python standard library; your code never leaves the machine and no model is called.

Two report tiers cover different needs. The free preview is fully offline with no call limit, giving you the total risk count, type distribution, and a one-line summary. The full report adds exact line numbers, per-item fix suggestions, and a server RSA-signed audit certificate (certified.json / certified.md) that can be archived, handed to a client, or verified offline at any later date. The tool states technical facts, not legal opinions; the final call is yours.

It suits developers who want to confirm code ownership after integrating open-source pieces, validate that a clean-room rewrite truly severed the upstream link, run a pre-delivery self-audit to avoid claiming derivative work as proprietary, or verify the provenance of external pull requests before merging.