vet-review

What problem does it solve?

Vetting PR review findings is hard because automated tools often produce noisy, low-signal suggestions; this Skill helps you validate which findings truly matter so engineers don’t waste time.

Core Features & Use Cases

• Skeptical validation of existing findings: evaluates prior review output (from /review-pr or a similar sub-agent) and drops anything that doesn’t survive real-world scrutiny.
• Context-aware assessment with surrounding code: reads the full file (or the changed sections with generous context) and checks invariants and call paths before deciding.
• Supports interactive and batch workflows: interactive mode walks finding-by-finding; batch mode emits a single JSON object with survived vs dropped items.

Use case: You receive a long automated PR review with many suggestions. Use vet-review to keep only the findings that represent actual bugs, security issues, logic gaps, or genuinely clarifying improvements.

Quick Start

Run vet-review after your existing review by invoking it with the PR number and batch mode, like: “Call vet-review for PR 123 with --batch to get a JSON list of survived and dropped findings.”