feat(evals): build and publish trace-derived evals locally - #476

Merged

aamir merged 11 commits into

mainfrom

codex/agent-authored-workload-evals

Aug 31, 2026

Commits on Aug 31, 2026

feat(evals): materialize complete workload weeks (U1)
aamir
committed
6a762b1
feat(evals): author and verify workload evals locally (U2)
aamir
committed
4087e7b
feat(evals): preview and publish eval releases (U3)
aamir
committed
86a344f
Address eval workflow review findings (#476)
aamir
committed
7be6ece
Bind eval releases to verified source exports (#476)
aamir
committed
5ac430e
Preserve complete frozen eval windows (#476)
aamir
committed
4705607
Bind eval releases to the exact source corpus (#476)
aamir
committed
4c49fa8
Require frozen time for hosted trace compilation (#476)
aamir
committed
3ec1c43
Fail closed on unsafe or empty eval builds (#476)
aamir
committed
0ecd4c2
Reject ambiguous eval bundle paths (#476)
aamir
committed
8b19cd2
feat(evals): support provisional draft checks
aamir
committed
4784487