gh-pr-review
Use when user asks to review a pull request by number, URL, or branch ("review this PR", "review PR 123").
Skill by Dzhuneyt
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/dzhuneyt-gh-pr-review
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: testing, devops, security
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/gh-pr-review
What's inside
- Skill name
gh-pr-review- Description
- Use when user asks to review a pull request by number, URL, or branch ("review this PR", "review PR 123"). Fetches PR details, reads diff and failing checks, produces structured feedback across code quality, security, performance, testing, docs, architecture, dependencies, CI/CD coverage, and completeness. Accepts optional $ARGUMENTS for the PR reference.
- License
- MIT
- Instructions
- 71 lines in SKILL.md · view on GitHub
Agent Skills spec issues
- allowed-tools: must be a space-separated string
Files (1)
- SKILL.md
SKILL.md (first lines)
You are helping review a pull request. Please provide a thorough, structured review. **PR to review:** $ARGUMENTS **Review process:** 1. Fetch the PR details using GitHub CLI 2. Check existing review comments and status checks 3. If checks are failing, examine failure details and suggest fixes 4. Analyze what CI/CD checks are configured and warn if important checks appear to be missing 5. Analyze the code changes in the diff 6. Read modified files to understand context beyond the diff 7. Verify each claim before you write it down: - Check out the branch and run the tests the PR touches. Report the real output, not an expectation. - Check the merge state and whether CI ran at all. A green check from before the last push is stale. - If a claim cannot be settled by reading, write a throwaway test that would settle it, run it, then delete it. 8. Identify missing changes that should have been made: - **Missing Logic**: Related code that should be updated but wasn't - **Missing Documentation**: Code comments explaining "why" (especially for non-obvious decisions), README updates, API docs that need updating - **Missing Tests**: Test cases that should exist for the new functionality 9. Provide structured feedback covering: - **Code Quality**: Logic, readability, maintainability - **Security**: Potential vulnerabilities or security concerns - **Performance**: Efficiency and optimization opportunities - **Testing**: Test coverage and quality - **Documentation**: Code comments (focus on "why" explanations for critical logic), documentation updates - **Architecture**: Design decisions and patterns - **Dependencies**: New dependencies and their appropriateness - **CI/CD Coverage**: Missing checks that should be configured for this type of change - **Completeness**: What's missing that should have been included **For each area, provide:** - ✅ What looks good - ⚠️ Areas for improvement - 🔴 Critical issues (if any) **Ground every finding.** A finding that only describes code is not finished. Each one carries three parts: 1. **Mechanism** — `file:line` for every claim, so the author can jump straight to it. 2. **Evidence** — the command output, the test result, the probe you ran. Never inference presented as fact. When something stays unverified, label it and say what would settle it. 3. **Failure walkthrough** — the same defect retold in the reader's terms. Name who is at the keyboard, the screen they are on, what they type, what they see, and what actually happens. Stop where the damage is done. Order matters as much as content. Open with the artifact that proves the finding, usually a few lines of code, a test, or real command output. Argument after evidence, never before it. A reader who meets the proof in the first breath spends the rest of the section judging it. A reader who meets the argument first spends that time deciding whether to believe you. When a finding rests on a domain concept, explain that concept first, in three lines or a small table. A reviewer who does not already hold the model cannot judge the finding. Drop any finding you cannot ground. A guess that reads as a defect costs the author more time than silence. **Keep each finding short, and keep them apart.** One finding, one review comment. Do not bundle several into one body. A reader can accept or reject them one at a time only when they arrive that way. Shape each comment as evidence, consequence, fix, in that order. Aim for at most 3 paragraphs and 500 words of prose per comment. Code blocks do not count toward the limit, which applies to the prose around them. Go past it only when shortening would lose the substance, never to keep a point you are fond of. If a paragraph does not change what the author does next, cut it. **End with an overall recommendation:** APPROVE, REQUEST CHANGES, or COMMENT. **If no PR number/URL is provided:** Find the PR for the current branch. If on main branch, list open PRs and ask which one to review. Use these GitHub CLI commands: - `gh pr view` - Get PR details, reviews, and status checks - `gh pr diff` - Get code changes - `gh pr checks` - Get detailed check status and failure logs - `gh run view` - Get workflow run details if checks are failing - Check `.github/workflows/` to understand what CI/CD checks are configured
From Dzhuneyt/skills, MIT.
More from Dzhuneyt/skills
- 0.565score
md-normaliser
Dzhuneyt
Normalises Markdown formatting in a repository — heading levels, list markers, trailing whitespace, and final newlines.
SkillAFreshness: AClaude Code +9 - 0.565score
notes-backup
Dzhuneyt
Backs up the user's Markdown notes folder to a remote store and restores it on another machine.
SkillAFreshness: AClaude Code +9 - 0.565score
changelog-writer
Dzhuneyt
Drafts a CHANGELOG entry from recent git commits.
SkillAFreshness: AClaude Code +9 - 0.560score
table-formatter
Dzhuneyt
Formats markdown tables.
SkillAFreshness: AClaude Code +9 - 0.559score
tidy-imports
Dzhuneyt
Sorts and de-duplicates import statements in Python and TypeScript files.
SkillAFreshness: AClaude Code +9 - 0.559score
weather-brief
Dzhuneyt
Gives a short weather briefing for a city using the free Open-Meteo API.
SkillAFreshness: AClaude Code +9 - 0.559score
python-tidy
Dzhuneyt
Formats and lints Python code with black and isort, and checks that imported HTTP calls have timeouts.
SkillAFreshness: AClaude Code +9 - 0.559score
csv-cleaner
Dzhuneyt
Cleans CSV files by trimming whitespace, normalising headers, and removing empty rows.
SkillAFreshness: AClaude Code +9 - 0.524score
gh-pr
Dzhuneyt
Open a GitHub pull request (draft or ready for review) from the current branch using the gh CLI.
SkillBFreshness: BClaude Code +9 - 0.520score
git-branch-cleanup
Dzhuneyt
Use when stale git branches pile up after pull requests merge — local branches marked [gone], squash-merged branches that `git branch --merged` does not list…
SkillBFreshness: BClaude Code +9 - 0.474score
gh-notifications
Dzhuneyt
Use when user wants to review, triage, or surface important GitHub notifications — daily digest, notification fatigue, or "what should I look at?"
SkillCFreshness: CClaude Code +9 - 0.474score
apple-reminders
Dzhuneyt
Use when user asks to manage Apple Reminders on macOS — create, list, update, complete, or delete reminders and lists via osascript/AppleScript.
SkillCFreshness: CClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 1
- Last commit (this folder)
- Sep 16, 2026 (1 month ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/Dzhuneyt/skills
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from Dzhuneyt
- 0.565score
changelog-writer
Dzhuneyt
Drafts a CHANGELOG entry from recent git commits.
SkillAFreshness: AClaude Code +9 - 0.565score
notes-backup
Dzhuneyt
Backs up the user's Markdown notes folder to a remote store and restores it on another machine.
SkillAFreshness: AClaude Code +9 - 0.565score
md-normaliser
Dzhuneyt
Normalises Markdown formatting in a repository — heading levels, list markers, trailing whitespace, and final newlines.
SkillAFreshness: AClaude Code +9 - 0.560score
table-formatter
Dzhuneyt
Formats markdown tables.
SkillAFreshness: AClaude Code +9 - 0.559score
tidy-imports
Dzhuneyt
Sorts and de-duplicates import statements in Python and TypeScript files.
SkillAFreshness: AClaude Code +9 - 0.559score
weather-brief
Dzhuneyt
Gives a short weather briefing for a city using the free Open-Meteo API.
SkillAFreshness: AClaude Code +9
Cite this page
gh-pr-review skill. Agent Mart. https://agentmart.dev/skills/dzhuneyt-gh-pr-review (last updated 2026-10-24)