What problem does it solve? Code reviews often degrade into opinion-based vibe checks that miss name shadowing, scope ambiguity, and regressions. This Skill turns post-implementation review into a structured certificate — premises, claims, evidence, and a verdict derived from explicit checkpoints — so correctness decisions are defensible rather than asserted. ## Core Features & Use Cases - Certificate-based review: Runs a fixed 6-step procedure (premises, function resolution, execution trace, regression check, edge cases, verdict) on every diff, producing a CORRECT / LIKELY_CORRECT / CONCERNS / BUGGY verdict with confidence and summary. - Spec-builder correctness gate: Acts as the mandatory correctness gate after a spec-builder task is implemented, run by an agent other than the one that wrote the code, complementing the validate-done-certificate completeness gate. - Skip conditions for trivial changes: Recognizes docs, formatting, and version-bump changes and records the gate as skipped rather than fabricating ceremony. - Use Case: A spec-builder sub-agent finishes implementing a passphrase-lock task; the orchestrator dispatches this Skill against the workspace diff, which resolves each call through the 5-step sequence, flags a shadowed verify() function, traces a concrete input, checks a downstream caller, and returns VERDICT: BUGGY with the offending line. ## Quick Start Ask the agent to run a semi-formal review of the diff produced by the just-completed task before marking it done.
