You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository was archived by the owner on Nov 2, 2026. It is now read-only.
Repository navigation
This repository was archived by the owner on Nov 2, 2026. It is now read-only.
Map reliable model testing and reviewer evidence in ProofShot #8
Produce an approved, implementation-ready workflow for model-led UI verification grounded in real agent-browser exploration, distilled into concise human testing guidance and fresh reviewer evidence, recoverable under interruption and concurrency, and assigned to the correct owner without making ProofShot the behavioral-testing authority.
Define the reviewer evidence and human testing handoff requires high-level human instructions, one concise replay per distinct use case, explicit assertions, curated screenshots, canonical verdicts, action disclosure, and private raw evidence.
All decision and implementation sub-issues are complete. The replay contract, managed native runtime, secure projection, and concurrent recording finalization fix are deployed; the final managed soak produced two clean PASS sessions with an empty registry afterward.
Activity
alvarohulse commentedon Aug 19, 2026
All decision and implementation sub-issues are complete. The replay contract, managed native runtime, secure projection, and concurrent recording finalization fix are deployed; the final managed soak produced two clean PASS sessions with an empty registry afterward.