Pr review

Skill Babalsaab/Project_Starter/skills/pr-review

Use when reviewing pull requests, checking code quality in diffs, running security analysis on changed files, or generating structured PR review reports.From its SKILL.md

Install

npx -y skills add Babalsaab/Project_Starter --skill pr-review

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

4.1 KB, as published. Nobody here has run it

PR Review Skill

Automated pull request quality analysis and structured review for the Vibecoded Orchestrator. This skill provides agents with review checklists, diff categorization, and a standardized output format so that every PR is evaluated consistently across security, quality, architecture, testing, and documentation concerns.

Review Categories

The skill evaluates PRs across five categories, each with its own checklist:

Security

No hardcoded secrets, API keys, or tokens
No SQL injection vectors
Input validation on user-facing endpoints
No eval() or dynamic code execution
Proper authentication/authorization checks

Code Quality

TypeScript strict mode compliance
No any types without justification
Functions under 50 lines
Files under 600 lines
Meaningful variable and function names
No dead code or commented-out blocks

Architecture

Changes follow existing patterns
No circular dependencies introduced
Provider abstraction used (not direct API calls)
Agent token caps respected
Configuration via globals.yaml, not hardcoded

Testing

New functionality has corresponding tests
Edge cases covered
No test-only code in production files

Documentation

README updated for new features
CLAUDE.md updated if agent/config changes
JSDoc on public functions
Inline comments for complex logic

Inputs

Input	Type	Description
`diff`	`string`	The full PR diff
`description`	`string`	The PR description / body text
`changed_files`	`string[]`	List of file paths changed in the PR
`pr_number`	`number` (optional)	PR number for reference
`author`	`string` (optional)	PR author for context

File Categorization

Changed files are automatically categorized into groups for focused review:

orchestrator -- files under orchestrator/ (excluding providers)
providers -- files under orchestrator/providers/
agents -- files under agents/
web-app -- files under src/
config -- files under config/
scripts -- files under scripts/
docs -- markdown files
other -- everything else

Output Format

The skill produces a structured PRReviewOutput with the following fields:

Verdict

One of three values:

APPROVE -- PR meets all quality standards and is safe to merge
REQUEST_CHANGES -- PR has issues that must be addressed before merging
COMMENT -- PR needs discussion but is not necessarily blocked

Summary

A 2-3 sentence overview of the PR's purpose and overall quality assessment (capped at 500 characters).

Issues

A list of problems found, each with:

Field	Type	Description
`severity`	`"critical" \| "warning" \| "info"`	Issue severity level
`file`	`string`	File where the issue was found
`line`	`number` (optional)	Line number if applicable
`message`	`string`	Description of the issue
`category`	`string`	Review category (security, quality, etc.)

Checklist

A list of checklist items with pass/fail status:

Field	Type	Description
`label`	`string`	The checklist item text
`passed`	`boolean`	Whether the check passed
`notes`	`string` (optional)	Additional context

Suggestions

A list of actionable improvement recommendations as plain strings.

Usage

This skill is injected into agent context during PR review tasks. The buildPRReviewContext function generates the review prompt with the diff, file categorization, and full checklist. The parseReviewOutput function then parses the agent's free-text response back into the structured PRReviewOutput format.