multi-agent-mcp
Allows use of the multi-agent MCP server with GitHub Copilot Chat in VS Code for automated software development workflows including planning, architecture, implementation, QA, and code review.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@multi-agent-mcp Execute full development workflow for a JWT authentication module in Node.js"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Multi-Agent MCP Server
A sophisticated Model Context Protocol (MCP) server that orchestrates multiple AI agents to collaborate on software development projects. This system enables automated, multi-agent workflows for planning, architecture design, implementation, quality assurance, and code review.
🚀 Features
Multi-Agent Collaboration
Product Manager: Analyzes requirements, creates user stories, prioritizes features
Software Architect: Designs system architecture, APIs, and data models
Software Engineer: Implements features with clean, efficient code
QA Engineer: Reviews code for bugs, edge cases, and vulnerabilities
Code Reviewer: Suggests improvements for readability and maintainability
MCP Tools
plan_feature- Generate detailed feature plans with user storiesdesign_architecture- Create system architecture and technical specificationsimplement_code- Generate production-ready code implementationsreview_code- Perform quality assurance and testing reviewssuggest_improvements- Provide code refactoring and optimization suggestionscollaborate- Enable multi-agent discussions and consensus buildingfull_workflow- Execute complete development lifecycle from planning to review
Automated Workflows
Full Development Cycle: Planning → Architecture → Implementation → QA → Review
Terminal Automation: Auto-executes build, test, and deployment commands
Conversation History: Maintains context across multi-agent interactions
Error Handling: Robust error recovery and retry mechanisms
Related MCP server: all-agents-mcp
📋 Prerequisites
Node.js 18+ and npm
TypeScript 5.3+
VS Code with GitHub Copilot Chat extension
Claude Desktop (optional, for alternative MCP client)
🛠️ Installation
Clone the repository
git clone <repository-url> cd multi-agent-mcpInstall dependencies
npm installBuild the project
npm run build
⚙️ Configuration
VS Code Settings (Recommended)
For full automation without terminal prompts, add these settings to your VS Code User Settings:
{
"chat.tools.terminal.autoApprove": {
"/.*/": true
},
"chat.mcp.autostart": "newAndOutdated"
}Claude Desktop Configuration
Add to your claude_desktop_config.json:
{
"mcpServers": {
"multi-agent": {
"command": "node",
"args": ["C:\\path\\to\\multi-agent-mcp\\build\\index.js"]
}
}
}🚀 Usage
Starting the Server
npm startUsing with GitHub Copilot
Open VS Code with GitHub Copilot Chat
Start the MCP server in a terminal:
npm startUse the full workflow tool:
Use the full_workflow tool with:
{
"requirement": "Create a user authentication system with JWT tokens",
"language": "typescript"
}Individual Agent Tools
Plan a Feature:
Use plan_feature with:
{
"requirement": "Build a REST API for user management"
}Design Architecture:
Use design_architecture with:
{
"feature_plan": "User management API with CRUD operations",
"tech_stack": "Node.js, Express, PostgreSQL"
}Implement Code:
Use implement_code with:
{
"architecture": "REST API with Express router pattern",
"language": "typescript"
}Review Code:
Use review_code with:
{
"code": "your code here",
"context": "User authentication module"
}Multi-Agent Collaboration:
Use collaborate with:
{
"topic": "Database schema design for e-commerce platform",
"agents": "architect,engineer",
"rounds": 3
}📁 Project Structure
multi-agent-mcp/
├── src/
│ └── index.ts # Main MCP server implementation
├── build/ # Compiled JavaScript output
├── package.json # Dependencies and scripts
├── tsconfig.json # TypeScript configuration
├── OPTIMIZATIONS.md # Performance optimizations
└── README.md # This file🔧 Development
Building
npm run buildWatch Mode
npm run watchTesting
npm test🤖 Agent Capabilities
Product Manager Agent
Requirements analysis and prioritization
User story creation with acceptance criteria
Feature planning and roadmap development
Success metrics definition
Software Architect Agent
System architecture design
Component and API specification
Data model design
Technology stack recommendations
Scalability and performance considerations
Software Engineer Agent
Clean code implementation
Algorithm optimization
Error handling and edge cases
Best practices adherence
Documentation generation
QA Engineer Agent
Code quality assessment
Bug detection and vulnerability scanning
Test case recommendations
Edge case identification
Performance bottleneck analysis
Code Reviewer Agent
Code readability evaluation
Maintainability improvements
Refactoring suggestions
Best practices validation
Performance optimizations
🔄 Workflow Example
Input: "Create a task management web app"
1. PM Agent → Analyzes requirements, creates user stories
2. Architect Agent → Designs React + Node.js architecture
3. Engineer Agent → Implements components and API endpoints
4. QA Agent → Reviews code, suggests test cases
5. Reviewer Agent → Recommends code improvements
Output: Complete, production-ready application🛡️ Security & Best Practices
Terminal Command Safety: Configurable auto-approval for development commands
Error Recovery: Robust error handling with retry mechanisms
Conversation Context: Maintains state across multi-agent interactions
Type Safety: Full TypeScript implementation with strict typing
🤝 Contributing
Fork the repository
Create a feature branch:
git checkout -b feature/new-agentMake your changes and test thoroughly
Submit a pull request with detailed description
📄 License
This project is licensed under the MIT License - see the LICENSE file for details.
🆘 Troubleshooting
Server Won't Start
Ensure Node.js 18+ is installed:
node --versionCheck dependencies:
npm installVerify build:
npm run build
MCP Tools Not Available
Restart VS Code after configuration changes
Check VS Code settings for MCP configuration
Verify server is running:
npm start
Terminal Commands Not Auto-Executing
Update VS Code settings with terminal auto-approve rules
Restart VS Code to apply settings changes
📊 Performance
Response Time: < 2 seconds for typical requests
Concurrent Agents: Supports multiple simultaneous workflows
Memory Usage: Optimized for long-running sessions
Error Recovery: Automatic retry with exponential backoff
🔮 Future Enhancements
Additional specialized agents (DevOps, Security, UX/UI)
Integration with external APIs and services
Custom agent training and fine-tuning
Workflow templates and presets
Real-time collaboration features
Plugin system for extensibility
Built with ❤️ using the Model Context Protocol
Available Tools
13 toolsanalyze_codebaseC
Analyze existing codebase structure, patterns, and quality metrics
Inputs: path, exclude_patterns, include_patterns.
clear_historyB
Clear the current session's conversation history
Inputs: no named parameters.
collaborateC
Multi-agent collaboration where agents discuss a topic and reach consensus through multiple rounds
Inputs: topic (required), agents (required), rounds.
design_architectureC
Architect designs the system architecture, data models, APIs, and scalability approach
Inputs: feature_plan (required), tech_stack, constraints.
estimate_effortC
Product Manager estimates story points and development time for a requirement
Inputs: requirement (required), complexity.
full_workflowB
Execute complete development workflow: planning → architecture → implementation → QA → review
Inputs: requirement (required), language, include_tests.
generate_testsC
Engineer generates comprehensive test suite for the provided code
Inputs: code (required), test_framework, coverage_target.
get_conversation_historyC
Retrieve the multi-agent conversation history with optional filtering
Inputs: last_n, agent_filter.
implement_codeC
Engineer implements the feature with clean, tested, production-ready code
Inputs: architecture (required), language, file_path, test_coverage.
list_sessionsB
List all available conversation sessions
Inputs: no named parameters.
plan_featureC
Product Manager analyzes requirements and creates a detailed feature plan with user stories, acceptance criteria, and success metrics
Inputs: requirement (required), context, priority.
review_codeC
QA Engineer performs comprehensive code review for bugs, security, edge cases, and test coverage
Inputs: code (required), context, focus_areas.
suggest_improvementsB
Code Reviewer suggests improvements for readability, maintainability, and performance
Inputs: code (required), focus_areas, performance_critical.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 13 tools
Most tools target distinct roles and phases, but review_code and suggest_improvements both analyze code for quality, and analyze_codebase overlaps with review activities. The boundaries are clear enough that agents can select correctly with minor caution.
Strong verb_noun pattern for most tools (plan_feature, design_architecture, implement_code, etc.), but collaborate is just a verb and full_workflow uses an adjective_noun form. The outliers are readable and not confusing, just less consistent.
13 tools is well-scoped for a multi-agent development workflow covering planning, architecture, implementation, review, testing, estimation, collaboration, and session management. Each tool clearly earns its place.
The surface covers the core development lifecycle end-to-end, including a full_workflow orchestrator and session history management. Minor gaps exist around explicit session creation or non-destructive session deletion, but agents can work around them.
Related MCP Connectors
Coordinate coding agents through MCP using existing AI plans, saved work, and independent checks.
AI work orchestration for plans, tasks, teams, and coding-agent dispatch.
Project management shared by people and AI agents, with persistent project state through MCP.
Command center for developers coordinating AI coding agents across projects.
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceA reusable AI software development team built on MCP. 13 specialized agents (Project Manager, Backend, Frontend, QA, Security, DevOps, UX, and more) collaborate via shared SQLite state. Exposes 44 MCP tools across 12 domains. All discussions and decisions are stored in the database so agents get back up to speed immediately when re-loaded.MIT
- AlicenseBqualityFmaintenanceEnables orchestrating multiple AI CLI agents (Claude Code, Codex, Gemini CLI, Copilot CLI) through a unified MCP interface for task delegation, cross-agent comparison, and specialized tools like code review and debugging.1411 npm13MIT
- FlicenseNot gradedqualityDmaintenanceEnables users to generate complete, production-ready software projects from simple ideas by coordinating 8 specialized AI agents through the Model Context Protocol.4-
- AlicenseNot gradedqualityDmaintenanceTransforms AI agents into spec-driven product engineers by managing the software project lifecycle through requirements, design, implementation, and archiving phases with state-aware MCP tools.5 npmMIT