KS
Killer-Skills

review-pr — how to use review-pr how to use review-pr, review-pr setup guide, ArcticDB code review automation, review-pr alternative, review-pr vs GitHub code review, what is review-pr, review-pr install, automating code reviews for data science

v1.0.0
GitHub

About this Skill

Perfect for Code Review Agents needing automated analysis of ArcticDB pull requests in the Python Data Science ecosystem. review-pr is a skill that performs thorough code reviews of ArcticDB pull requests, utilizing context variables like PR_NUMBER and REPO.

Features

Resolves PR number and repository using injected context variables
Supports invocation from GitHub Actions workflow or as a local slash command
Detects PR number from the current branch if $ARGUMENTS is empty
Utilizes bash commands for automation
Integrates with ArcticDB, a high-performance, serverless DataFrame database

# Core Topics

man-group man-group
[2.2k]
[165]
Updated: 2/27/2026

Quality Score

Top 5%
55
Excellent
Based on code quality & docs
Installation
SYS Universal Install (Auto-Detect)
Cursor IDE Windsurf IDE VS Code IDE
> npx killer-skills add man-group/ArcticDB/review-pr

Agent Capability Analysis

The review-pr MCP Server by man-group is an open-source Categories.community integration for Claude and other AI agents, enabling seamless task automation and capability expansion. Optimized for how to use review-pr, review-pr setup guide, ArcticDB code review automation.

Ideal Agent Persona

Perfect for Code Review Agents needing automated analysis of ArcticDB pull requests in the Python Data Science ecosystem.

Core Value

Empowers agents to perform thorough code reviews of ArcticDB pull requests using GitHub Actions workflow variables like PR_NUMBER, REPO, and EVENT_ACTION, ensuring high-quality code submissions.

Capabilities Granted for review-pr MCP Server

Automating code reviews for ArcticDB pull requests
Analyzing code changes between BEFORE_SHA and AFTER_SHA commits
Detecting PR numbers from current branch or $ARGUMENTS

! Prerequisites & Limits

  • Requires GitHub Actions workflow setup
  • Limited to ArcticDB and Python Data Science ecosystem
  • Needs $ARGUMENTS or PR_NUMBER to be set
SKILL.md
Readonly

Perform a thorough code review of an ArcticDB pull request.

Context variables

When invoked from a GitHub Actions workflow the following variables are injected before this prompt: PR_NUMBER, REPO, EVENT_ACTION, BEFORE_SHA, AFTER_SHA. When invoked locally as a slash command the PR number is $ARGUMENTS (e.g. /review-pr 2933). If $ARGUMENTS is empty, detect from the current branch.


Step 1 — Resolve PR number and repository

bash
1# If PR_NUMBER is not set and $ARGUMENTS is empty, detect from current branch: 2gh pr view --json number -q .number 3 4# If REPO is not set, detect: 5gh repo view --json nameWithOwner -q .nameWithOwner

Step 2 — Fetch diffs

If EVENT_ACTION is synchronize and both BEFORE_SHA and AFTER_SHA are set:

Attempt to produce the delta diff (new commits only):

bash
1git fetch --depth=1 origin <BEFORE_SHA> 2git diff <BEFORE_SHA>..<AFTER_SHA>
  • If the delta diff is non-empty, use it as the delta diff. Do not fetch the full diff — it is not needed (the existing sticky summary covers the earlier commits).
  • If the fetch fails (e.g. force-push removed BEFORE_SHA) or the result is empty, fall back: fetch the full diff and use it for both inline comments and summary.
bash
1gh pr diff <PR_NUMBER> # fallback only

Otherwise (event is opened, reopened, ready_for_review, or this is a local run):

Fetch the full diff and use it as both the full diff and the delta diff:

bash
1gh pr diff <PR_NUMBER>

Step 3 — Fetch existing inline review comments and sticky summary

Fetch all existing inline review comments (for deduplication):

bash
1gh api --paginate "repos/<REPO>/pulls/<PR_NUMBER>/comments" \ 2 --jq '[.[] | {path: .path, line: (.line // .original_line), author: .user.login, body: (.body | .[0:300])}]'

Also fetch the existing sticky summary comment, if any (for incremental updates):

bash
1gh api "repos/<REPO>/issues/<PR_NUMBER>/comments" \ 2 --jq '[.[] | select(.body | startswith("## ArcticDB Code Review Summary"))] | last | {id: .id, body: .body}'

Step 4 — Deduplication rule

Before posting any inline comment, check the existing comments for an entry with the same path and line:

  • Human comment exists: skip the inline comment. Note the finding in the summary checklist if relevant.
  • Previous Claude comment exists: skip the inline comment. Ensure the finding is still reflected in the summary checklist.
  • No existing comment: post the inline comment normally.

Step 5 — Review outputs

Inline comments

Post only on lines present in the delta diff, subject to the deduplication rule above.

Summary checklist

If EVENT_ACTION is synchronize and a previous sticky summary exists:

  • Evaluate only the delta diff against the guidelines.
  • Read the previous summary checklist.
  • Update individual checklist items that are affected by the delta:
    • A new commit that introduces an issue: change the item to ❌ and add a note.
    • A new commit that fixes a previously flagged issue: change the item to ✅ and note the fix.
    • Items unaffected by the delta: leave unchanged.
  • Post the amended summary as the updated sticky comment.

Otherwise (event is opened, reopened, ready_for_review, or no previous summary):

Evaluate the full diff and post a fresh summary checklist.


Step 6 — Load review guidelines

Read the file docs/claude/PR_REVIEW_GUIDELINES.md from the repository root using the Read tool. Apply all sections relevant to the files changed in the PR. The file also contains the summary checklist format to use when posting the sticky comment.

Related Skills

Looking for an alternative to review-pr or building a Categories.community AI Agent? Explore these related open-source MCP Servers.

View All

widget-generator

Logo of f
f

widget-generator is an open-source AI agent skill for creating widget plugins that are injected into prompt feeds on prompts.chat. It supports two rendering modes: standard prompt widgets using default PromptCard styling and custom render widgets built as full React components.

149.6k
0
Design

chat-sdk

Logo of lobehub
lobehub

chat-sdk is a unified TypeScript SDK for building chat bots across multiple platforms, providing a single interface for deploying bot logic.

73.0k
0
Communication

zustand

Logo of lobehub
lobehub

The ultimate space for work and life — to find, build, and collaborate with agent teammates that grow with you. We are taking agent harness to the next level — enabling multi-agent collaboration, effortless agent team design, and introducing agents as the unit of work interaction.

72.8k
0
Communication

data-fetching

Logo of lobehub
lobehub

The ultimate space for work and life — to find, build, and collaborate with agent teammates that grow with you. We are taking agent harness to the next level — enabling multi-agent collaboration, effortless agent team design, and introducing agents as the unit of work interaction.

72.8k
0
Communication