MCP PR Review Pipeline

MCP PR Review Pipeline

Multi-agent PR review: fetch via MCP, three parallel review agents, human gate, post via MCP

PR Data Fetch

Retrieves pull request metadata, diff, and changed files from GitHub via MCP

Quality Review

Evaluates code quality — logic, architecture, testing, error handling, and maintainability

Review Gate

Human gate — pauses for you to review findings before posting to GitHub

Review Posting

Posts the approved review to GitHub via MCP — summary comment, inline comments, and verdict

Security Review

Scans changed files for security vulnerabilities, credential exposure, and unsafe patterns

Style Review

Checks code style, naming conventions, formatting, and project-specific patterns

Fetch PR Data

Retrieves and structures pull request data from GitHub via MCP

Post Review

Formats and posts the approved review to GitHub via MCP

Quality Analysis

Evaluates code quality — logic, architecture, testing, and maintainability

Review Gate Summary

Synthesises findings from all review agents into a human-readable gate summary

Security Analysis

Analyses PR diff for security vulnerabilities and credential exposure

Style Analysis

Checks code style, naming, formatting, and project-specific conventions

GitHub MCP

GitHub MCP server for pull request access, code browsing, and review posting

LLM Service

Language model service for code analysis, security scanning, and review generation