Tags: github/gh-aw
Tags
Add five trajectory graders and harden generator contracts (#64023) * Implement five trajectory graders Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com> * Implement graders ranked 18-22 Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com> * Use total ordering for tool-wise-score tool calls Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com> * Harden trajectory grader contracts against malformed/missing evidence Co-authored-by: gh-aw-bot <259018956+gh-aw-bot@users.noreply.github.com> * Split malformed-action and missing-consequential-flag messages Co-authored-by: gh-aw-bot <259018956+gh-aw-bot@users.noreply.github.com> * Make all trajectory graders built-in and run them on every trace Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com> * Report trajectory graders without IR as unavailable and scope size guard Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com> * Extract shared Trajectory IR candidate lookup helper Co-authored-by: gh-aw-bot <259018956+gh-aw-bot@users.noreply.github.com> * Cover non-record traces in trajectory grader tests Co-authored-by: gh-aw-bot <259018956+gh-aw-bot@users.noreply.github.com> --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com> Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: gh-aw-bot <259018956+gh-aw-bot@users.noreply.github.com>
Report gateway steering events in audit output (#62943) * Report gateway steering events in audits Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com> * Add ADR for gateway steering audit reporting * Fix gateway steering cache propagation Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com> --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com> Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Avoid redownloading cached runs during logs audit (#61871) * Fix cached logs audit reuse Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com> * Preserve cached audit comparison evidence Co-authored-by: gh-aw-bot <259018956+gh-aw-bot@users.noreply.github.com> * Clarify cached JSONL audit reuse gate Co-authored-by: gh-aw-bot <259018956+gh-aw-bot@users.noreply.github.com> --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com> Co-authored-by: gh-aw-bot <259018956+gh-aw-bot@users.noreply.github.com>
PreviousNext