MCP Agent Network CLI

A powerful command-line interface for interacting with the MCP (Model Communication Protocol) Agent Network - a microservices-based AI agent system.

Features

• 🤖 Model Router: Route prompts to appropriate AI models with intelligent routing
• 📝 Plan Management: Create and manage development plans and tasks
• 🌿 Git Worktree: Manage Git worktrees for isolated development
• ⚡ Workflow Orchestrator: Create and execute complex workflows
• 🔍 Verification Feedback: Submit feedback and verify code quality
• 🎯 Interactive Mode: User-friendly interactive interface
• 📊 System Status: Monitor service health and performance

Installation

Prerequisites

• Node.js 16.0.0 or higher
• MCP Agent Network services running (see main project README)

Install Dependencies

cd cli
npm install

Make CLI Globally Available

npm link

Or run directly:

chmod +x bin/mcp.js

Quick Start

Check System Status

mcp status
mcp health

Interactive Mode

mcp interactive
# or
mcp i

Model Router

# Ask AI a question
mcp model route "Explain microservices architecture"

# Force specific model
mcp model route "Write a Python function" --model sonnet

# List available models
mcp models

Plan Management

# Create a new plan
mcp plan create "Build user authentication" --description "Implement JWT-based auth system"

# List all plans
mcp plan list

# Show plan details
mcp plan show <plan-id>

# Update plan status
mcp plan update <plan-id> --status active

Git Worktree Management

# Create worktree
mcp git create https://github.com/user/repo.git feature-branch

# List worktrees
mcp git list

# Remove worktree
mcp git remove <worktree-id>

Workflow Orchestration

# Create workflow
mcp workflow create "Deploy Application" --steps '[{"name":"build","type":"code_generation"},{"name":"test","type":"verification"}]'

# Execute workflow
mcp workflow execute <workflow-id>

# Check workflow status
mcp workflow status <workflow-id>

Verification & Feedback

# Submit feedback
mcp verify submit bug_report "API endpoint returning 500" --severity high --description "User login endpoint fails intermittently"

# List feedback
mcp verify list

# Verify code file
mcp verify code src/auth.py --language python --types syntax,security,quality

Configuration

The CLI reads configuration from:

1. Environment variables
2. .env file in the project root
3. Command-line options

Environment Variables

# Service endpoints
MODEL_ROUTER_URL=http://localhost:8001
PLAN_MANAGEMENT_URL=http://localhost:8002
GIT_WORKTREE_URL=http://localhost:8003
WORKFLOW_ORCHESTRATOR_URL=http://localhost:8004
VERIFICATION_FEEDBACK_URL=http://localhost:8005

# Debug mode
DEBUG=true

Command Reference

Global Options

• --host <host>: API host (default: localhost)
• --verbose: Verbose output
• --json: Output in JSON format

System Commands

• mcp status: Check service status
• mcp health: Detailed health check

Model Router Commands

• mcp model route <prompt>: Route prompt to AI model
  • --model <model>: Force specific model
  • --temperature <temp>: Model temperature (default: 0.7)
• mcp model models: List available models

Plan Management Commands

• mcp plan create <title>: Create new plan
  • --description <desc>: Plan description
  • --tags <tags>: Comma-separated tags
• mcp plan list: List all plans
  • --status <status>: Filter by status
• mcp plan show <planId>: Show plan details
• mcp plan update <planId>: Update plan
  • --status <status>: Update status

Git Worktree Commands

• mcp git create <repo> <branch>: Create worktree
  • --path <path>: Custom worktree path
• mcp git list: List all worktrees
• mcp git remove <worktreeId>: Remove worktree

Workflow Orchestrator Commands

• mcp workflow create <name>: Create workflow
  • --steps <steps>: Workflow steps JSON
• mcp workflow execute <workflowId>: Execute workflow
• mcp workflow status <workflowId>: Check workflow status

Verification Feedback Commands

• mcp verify submit <type> <title>: Submit feedback
  • --description <desc>: Feedback description
  • --severity <severity>: Severity level (low|medium|high|critical)
• mcp verify list: List feedback entries

Development

Project Structure

cli/
├── bin/mcp.js              # CLI entry point
├── commands/               # Command handlers
│   ├── system.js
│   ├── model-router.js
│   ├── plan-management.js
│   ├── git-worktree.js
│   ├── workflow-orchestrator.js
│   ├── verification-feedback.js
│   └── interactive.js
├── lib/                    # Utilities and clients
│   ├── clients/            # Service clients
│   ├── http-client.js      # HTTP client utility
│   └── utils.js            # Helper functions
├── config/                 # Configuration
└── package.json

Adding New Commands

1. Create command handler in commands/
2. Create service client in lib/clients/
3. Register command in bin/mcp.js
4. Update README

Error Handling

The CLI includes comprehensive error handling:

• Network timeouts and retries
• Service unavailability detection
• Input validation
• Graceful error messages

Troubleshooting

Service Connection Issues

# Check if services are running
mcp status

# Test individual service health
curl http://localhost:8001/health

Permission Issues

# Make CLI executable
chmod +x bin/mcp.js

# Install globally
sudo npm link

Environment Configuration

# Copy example environment file
cp ../.env.example ../.env

# Edit configuration
nano ../.env

Contributing

1. Fork the repository
2. Create feature branch: git checkout -b feature/new-command
3. Make changes and test: npm test
4. Submit pull request

License

MIT License - see LICENSE file for details.