review-flow
Automated AI code reviews powered — webhook-driven, real-time dashboard, MCP integration, smart queue with deduplication, multi-agent audits, and iterative follow-up reviews for GitLab MRs and GitHub PRs
claude mcp add --transport stdio dgouron-review-flow node path/to/server.js \ --env PORT="Port for the MCP server to listen on" \ --env LOG_LEVEL="Logging level (e.g., info, debug)" \ --env CLAUDE_API_KEY="API key for Claude Code (if configured)" \ --env WEBHOOK_SECRET="Webhook secret used for validating incoming GitLab/GitHub webhooks"
How to use
Reviewflow runs an automated AI code review workflow that can be triggered from your GitLab or GitHub repositories. The MCP integration provided by the server allows Claude Code to interact with the review process through structured tools such as get_workflow, start_agent, complete_agent, set_phase, get_threads, and add_action. Use the dashboard to monitor live progress, view agent results, and follow up automatically when developers push fixes. This setup enables multi-agent reviews (e.g., Clean Architecture, Security, Testing) and provides a real-time, WebSocket-enabled dashboard to observe phase transitions, agent status, and review history. The server is designed to work with both GitLab and GitHub, using secure CLI-based OAuth for authentication instead of embedding API tokens.
To use Reviewflow with MCP, ensure your MCP server is running and reachable by webhook callbacks. The MCP tools allow Claude to report progress, advance review phases, fetch discussion threads, and queue thread actions such as resolving, replying, or commenting. When a developer pushes new changes that address blocking issues, Reviewflow can re-read threads, verify resolutions, and post follow-up summaries with updated scores, creating an iterative review loop.
How to install
Prerequisites:
- Node.js >= 20 installed on the host machine
- npm (comes with Node.js) or pnpm/yarn if preferred
- GitHub or GitLab project with webhooks support and access configured
Installation steps:
- Install the Reviewflow CLI globally (example):
npm install -g reviewflow
- Initialize or configure your project for MCP integration (interactive):
reviewflow init
- Follow prompts to configure server port, webhook secrets, and MCP integration with Claude Code.
- For non-interactive setup, use:
reviewflow init --yes
- Start the Reviewflow server:
reviewflow start
- The dashboard will be available at http://localhost:3847 by default (adjust via configuration).
- Validate your setup:
reviewflow validate
- If you already have a running MCP server, ensure its configuration matches the provided mcp_config and environment variables.
Additional notes
Tips and common issues:
- Ensure Node.js version matches the project requirements (Node.js >= 20).
- Keep WEBHOOK_SECRET secure; rotate periodically and update in both your MCP config and repository settings.
- If the MCP server cannot connect to Claude Code, check network permissions, API keys, and that the MCP port is accessible from the webhook delivery path.
- The dashboard uses WebSocket; ensure your environment allows WebSocket connections through firewalls.
- When updating config, restart the MCP server to apply changes.
- If you encounter credential issues with GitHub/GitLab, rely on the OAuth flow via CLI tools (gh/glab) as documented; no token in config files.
- Reviewflow’s agents are configurable; you can tailor the agent list to match your project standards by editing the agents block in the project configuration.
- For troubleshooting, consult the CLI reference and deployment guide linked in the Documentation section of the README.
Related MCP Servers
better-chatbot
Just a Better Chatbot. Powered by Agent & MCP & Workflows.
apple-mail
MCP server giving AI assistants full access to Apple Mail - read, search, compose, organize & analyze emails via natural language
hyperliquid
🚀 MCP Server para Hyperliquid DEX - Trade com Claude usando linguagem natural. Desenvolvido por Caio Vicentino para as comunidades Yield Hacker, Renda Cripto e Cultura Builder
architect
A powerful, self-extending MCP server for dynamic AI tool orchestration. Features sandboxed JS execution, capability-based security, automated rate limiting, marketplace integration, and a built-in monitoring dashboard. Built for the Model Context Protocol (MCP).
appstore-connect
A Model Context Protocol (MCP) server for Apple's App Store Connect API
bitbucket-automatic-pr-reviewer
🤖 Automated PR reviews using Claude CLI with Bitbucket webhooks. Features sequential processing, MCP integration, Prometheus metrics, and secure webhook validation. Perfect for teams wanting AI-powered code reviews without API costs.