review

Route and orchestrate PR reviews by applying routing rules from the concern registry.

19|31|Updated May 29, 2025

Search Tags:#routing #code-quality #risk-assessment #workflows #reviewers #pr-review

Ready to run

npx skills add https://github.com/grafana/grafana-pathfinder-app --skill review-grafana

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: review
Source: https://github.com/grafana/grafana-pathfinder-app/tree/main/.cursor/skills/review
Command: npx skills add https://github.com/grafana/grafana-pathfinder-app --skill review-grafana

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Route and orchestrate PR reviews by reading the concern registry and applying routing rules.

Core Features & Use Cases

Structured, multi-phase PR review workflow that classifies changes and routes concerns using a design registry.
Automated reviewer dispatch and parallel execution for security, correctness, and other always-on concerns.
Transparent outputs including activated_concerns, activation_reason, risk_signals, likely_one_way_doors, reviewers_to_run, and coverage_confidence.

Quick Start

Provide the PR changes so the skill can guide you through seven review phases end-to-end.

Frequently Asked Questions about review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I orchestrate PR reviews for code quality and risk assessment?▼

PR review orchestration works by reading a concern registry and applying routing rules to your code changes. It executes seven phases covering classification, routing, reviewer selection, and execution to ensure comprehensive coverage.

How does a concern registry route PR reviewers for different change classes?▼

A concern registry routes PR reviewers by applying predefined routing rules to classify diverse change classes. It activates concerns based on risk signals and selects appropriate reviewers for parallel execution across security, correctness, and other always-on areas.

What outputs can I expect from an automated PR review workflow?▼

An automated PR review workflow enforces structured outputs including activated_concerns, activation_reason, risk_signals, likely_one_way_doors, reviewers_to_run, and coverage_confidence. These provide transparent reasoning for every review decision.

Do I need design docs to run a multi-phase PR review?▼

Yes, you need design docs and routing tables in your repository, specifically located at docs/design/CONCERNS.md. The PR review workflow depends on these files to read the concern registry and apply the correct routing rules.

Can I automate reviewer dispatch for security and correctness concerns?▼

Yes, you can automate reviewer dispatch for security, correctness, and other always-on concerns. The PR review workflow handles this by routing concerns and executing selected reviewers in parallel based on your change classification.

What is the best way to assess one-way door risks in a pull request?▼

To assess one-way door risks in a pull request, use a structured review workflow that identifies risk signals and likely_one_way_doors. It evaluates these risks during the classification phase to guide reviewer selection and execution.