Evaluation Receipt
A credential file in AI code review workflows that records review results, including fields such as diff hash, model ID, and review conclusion. It is used by CI gates to determine whether a code change should be allowed to proceed.
Timeline (last 90 days)
When an AI agent is responsible for code review in a CI pipeline, reusing an old evaluation receipt after a model failover causes the gate to pass a change that the new model has never scored
An evaluation receipt must be bound to its corresponding diff, prompt, model, and workspace generation; a matching diff hash alone does not constitute a valid review
Hashing does not measure review quality; the gate only checks for the existence of identity and rationale, and a matching receipt may still describe a cursory reading of the diff
All Facts (3)
When an AI agent is responsible for code review in a CI pipeline, reusing an old evaluation receipt after a model failover causes the gate to pass a change that the new model has never scored
50%UnverifiedAn evaluation receipt must be bound to its corresponding diff, prompt, model, and workspace generation; a matching diff hash alone does not constitute a valid review
50%UnverifiedHashing does not measure review quality; the gate only checks for the existence of identity and rationale, and a matching receipt may still describe a cursory reading of the diff
50%