TAINT: on-chain proof of supply-chain injection, rewards without the human triage

Sharing an early-stage protocol design I’d like pushback on.

Existing supply-chain defenses verify that an artifact was built from intended blindly trusted components. They don’t answer the question:

can someone get arbitrary bytes into a published artifact?

TAINT rewards cryptographic proof of injection capability. A claimer commits a secret, the contract derives an unpredictable challenge, and the claimer must cause a derived flag to appear inside a scoped region of a published artifact — then prove its inclusion via Merkle proof.

A smart contract verifies and pays out automatically.

If someone can place a harmless flag, they could have placed malware; rewarding the harmless version reveals the capability before it gets abused.

Paper and repo: GitHub - juli/taint: TAINT: Crypto protocol for testing software supply chain integrity · GitHub

Looking for general feedback across the whole protocol: construction, grinding analysis under more permissive extraction strategies, entropy source trade-offs, and the reward mechanism (particularly control claims and propagation).

Prior art I may have missed also welcome; the closest neighbors I found were CHEESECLOTH and the attestation frameworks (signatures, reproducible builds, SLSA, in-toto, Sigstore) , but TAINT proves a different property than either.

Happy to dig into any piece in the thread.