What problem does it solve? Turning raw code review findings into a comment that reads like a human reviewer wrote it is tedious, and pasting severity tags, triage tables, and citation chains into an MR makes authors tune out. This Skill rewrites verified findings into concise prose, shows the draft for approval, and posts it to GitLab or GitHub. ## Core Features & Use Cases - Human-voiced rewriting: Strips scaffolding (severity labels, tables, verdict matrices) and leads each finding with user impact, one citation per claim, and a closing note on what is right. - Approval gate: Shows the drafted comment in chat and waits for approval before posting, unless the user passes --post up front. - Multi-platform posting: Posts via glab for GitLab, gh for GitHub, or a curl fallback against the GitLab API when glab is not authenticated, with optional line-anchored inline threads. - Use Case: After running an adversarial review on a merge request, say "post this as a review comment" to get a polished blocker-focused comment drafted, approved, and published with its note URL reported back. ## Quick Start Ask the assistant to post the current review findings as a comment on your merge request, for example by saying "post this as a review comment on MR 42".
