pr-comments_skill

This skill exports, preprocesses, analyzes, fixes, and resolves PR review comments to streamline code-review workflows and improve feedback quality.
  • Python

41

GitHub Stars

3

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill donghaozhang/video-agent-skill --skill pr-comments

  • review-batch.md2.3 KB
  • review-fix.md2.0 KB
  • SKILL.md3.9 KB

Overview

This skill exports GitHub PR review comments, preprocesses them for automated evaluation, and helps fix or reject individual review items. It organizes comments into task files, analyzes grouping and ordering by file/line, and supports single or batch fixes plus GitHub thread resolution. Use it to streamline review-followup from bots like CodeRabbit or Gemini.

How this skill works

The skill exports all review comments from a target PR into one markdown file per comment. A preprocess step cleans the content (removing noisy blocks and adding evaluation prompts) and produces task files. Analyze groups tasks by source file and recommends a bottom-up order. Fix evaluates a single task against the source file and either edits, explains, or marks it not applicable. Batch runs grouped fixes per file (descending line order) to avoid line-shift conflicts, and resolve closes threads on GitHub and archives completed tasks.

When to use it

  • You need a reproducible export of all PR review comments for analysis.
  • You want automated evaluation and suggested fixes for bot-generated reviews (CodeRabbit, Gemini).
  • You must apply multiple review suggestions safely without shifting line numbers.
  • You want to resolve PR review threads on GitHub after applying fixes.
  • You need a clear workflow to track pending vs completed review tasks.

Best practices

  • Install and authenticate GitHub CLI (gh) and ensure jq is available for JSON handling.
  • Export comments first, then run preprocess so tasks contain clean prompts for automated agents.
  • Use analyze to group by file and sort descending by line number to prevent line-shift conflicts.
  • Prefer batch mode for files with multiple comments; it applies fixes bottom-up per file.
  • When a fix is accepted or already present, run resolve to close the GitHub thread and move the task to completed.

Example use cases

  • Export a PR from a bot-heavy review to turn comments into actionable task files for evaluation.
  • Preprocess and batch-fix all CodeRabbit suggestions in one pass, grouped per file and applied bottom-up.
  • Analyze a large PR to find files with clustered comments and prioritize fixes where they matter most.
  • Fix an individual Gemini review comment by reading the source, applying a minimal code change, or explaining why it is not applicable.
  • Resolve a thread on GitHub automatically after confirming a fix and archive the task file.

FAQ

You need GitHub CLI (gh) authenticated and jq installed for JSON processing.

Will batch mode avoid breaking other suggestions?

Yes. Batch groups tasks by file and sorts fixes by descending line number so earlier edits do not shift later line numbers.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational