pr-review-fixer_skill

This skill automates PR feedback processing by validating issues, generating fixes, and verifying CI status across code-level and PR-level comments.

Python

16

GitHub Stars

1

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 arjenschwarz/agentic-coding --skill pr-review-fixer

SKILL.md8.6 KB

Overview

This skill automates fetching unresolved GitHub PR comments, validates actionable feedback, generates review overview and task files, and applies fixes. It also checks CI status and repairs failing tests, lint errors, type issues, and build failures, tracking iterations and preserving context.

How this skill works

It queries PR-level and code-level comments, filters out resolved or noisy items, and keeps only the latest claude[bot] comment per thread. Validated issues are written to a review-overview file and converted into runnable tasks in a review-fixes file; the skill then iterates through tasks to implement fixes, runs local checks, and updates CI status. Fixed code-level threads are resolved on GitHub while skipped or invalid items are left for human review.

When to use it

When addressing a pull request with multiple unresolved review comments.
When CI checks are failing after initial changes and need automated triage.
To consolidate duplicate review feedback into actionable tasks.
When you want iteration-tracked review overviews and fix task lists.
Before pushing follow-up commits to ensure tests and linters pass.

Best practices

Filter out resolved threads and non-actionable bot noise before creating tasks.
Keep only the latest claude[bot] comment per thread to avoid duplicate guidance.
Validate each issue against current code to skip stale or already-fixed feedback.
Run tests, linters, and type checks locally before committing fixes.
Increment iteration numbers in review files and preserve diff context in notes.

Example use cases

Automating fix work for a PR with dozens of code comments and a failing test suite.
Converting scattered reviewer suggestions into a prioritized task list for batch work.
Reconciling CLAUDE-generated suggestions with project conventions and applying fixes.
Resolving code-level review threads programmatically after applying validated fixes.
Tracking iterative review rounds by generating review-overview-[N].md and review-fixes-[N].md.

FAQ

Resolved threads, reviews with empty bodies, approval-only reviews, and non-actionable automated bot comments are skipped.

How does iteration tracking work?

The skill finds the highest existing iteration in the output directory and increments it, producing review-overview-[N].md and review-fixes-[N].md for each round.

Are all fixes applied automatically?

Validated issues are auto-fixed, but skipped or invalid items are left unresolved and require human attention.