domains / families / agent-tooling / task-review

task-review

separate

SkillsBench task PR review — classifies the task track (standard / research / multimodal), runs static policy checks against the track-specific rubric, benchmarks the task across oracle plus Claude and Codex (with and without skills), audits trajectories for cheating and skill invocation, and produces a `pr-N-task-timestamp-run.txt` review report alongside a `prN.zip` bundle of trajectories. Use when reviewing a SkillsBench task PR (by number, branch, or local task path), when the user asks to review a task, run benchmarks on a PR, audit a submission, classify a task as research or multimodal track, or prepare a comment to post on a SkillsBench PR.

When to use

• reviewing a SkillsBench task PR
• classifying a task as research or multimodal track
• running benchmarks on a PR
• auditing a submission
• preparing a comment to post on a SkillsBench PR

When not to use

• git checkout or git pull operations
• writing a verdict without running benchmarks
• interpreting results without auditing trajectories

Skill page

Source