What problem does it solve? When an agent produces an open-source prior-art search artifact β a keyword map, a per-angle search output, a per-repo extraction, or a synthesis report β someone must independently decide whether it is sound enough to feed the survey's downstream stages. Self-review by the producing agent approves its own work; this Skill applies a fixed, shared bar the reviewer cannot silently relax. ## Core Features & Use Cases - Twenty-two-condition acceptance gate: Judges keyword maps (conditions 1β6), search outputs (7β11), extractions (13β18), and synthesis reports (19β22), with proportionality (condition 12) gating every finding so thin-but-honest results pass. - Delegated deterministic checks: Runs the co-installed producer skill's validator (validate_prior_art.py) for schema and coverage conditions 7 and 11, requiring exit 0 and never re-implementing the checks. - Single machine-parseable verdict: Emits exactly one terminal VERDICT: approve|revise line (or the caller's named replacement), with revise findings naming the condition number, concrete gap, and what would satisfy it. - Use Case: After code-prior-art-survey produces a keyword map for a new capability, run this gate before search children execute it; on revise, route the condition-named findings back to the producer. ## Quick Start Use reviewing-code-prior-art-survey to judge the keyword map the producer just generated and give me the verdict with any findings.
