Repository navigation
fix(tracking): stop all-person runs on identity failures - #131
Merged
Merged
Conversation
austinjiann
changed the base branch from
codex/cache-recovery-receipt
to
main
November 21, 2026 04:17
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
An
--all-peoplerun could mark tracking successful after its identity audit rejected some tracks: the runner skipped that subset and continued with the rest. The fresh city baseline reproduced this with four of five tracks rejected, lamp false positives, and a fragmented main-walker track.Stop all-person runs on identity-audit failure. Also refuse avatar work against a retained failed audit, including individually passing slots, so an older successful stage record cannot bypass the guard. Explicit subset runs retain their existing skip behavior. Evidence and paid-attempt history are preserved.
Validation: 2,576 Python tests passed, including 12 new identity-gate cases; Ruff, formatting, and diff checks passed. A read-only replay against the actual city audit blocked both failed and individually passing slots without provider calls. The live tracking worker completed 94 samples with matching source indices, but visual and identity acceptance failed. This PR fixes acceptance, not tracking quality; no avatar or world was generated from the result. Stacked on #130.