Auto-Developer MCP Server

Model Context Protocol server providing AI-accessible tools for system control and multi-agent coordination.

🔗 Repository Info

• GitHub: https://github.com/[username]/auto-dev-mcp
• Docker Hub: ghcr.io/auto-developer/mcp
• Protocol: Model Context Protocol (MCP)

🎯 Purpose

• AI tool interface layer for LLMs (Claude, GPT-4, etc.)
• Multi-agent task coordination
• Structured tool execution with validation
• System state introspection for AI agents

📁 Structure

auto-dev-mcp/
├── src/                    # MCP server implementation
├── tests/                  # Unit and integration tests
├── schemas/                # JSON schemas for tool validation
├── examples/               # Usage examples for different LLMs
└── docs/                   # Tool documentation

🚀 Quick Start

pip install -r requirements.txt
python -m src.main

🔧 Available Tools

• System Tools: get_system_status, get_satellites, health_check
• Approval Tools: get_approval_queue, approve_request, deny_request
• Project Tools: create_project, deploy_project, list_templates
• Knowledge Tools: query_notes, search_web, update_research
• Canvas Tools: update_board, create_task, move_task

🤖 Agent Composer

Includes the interactive Agent Composer tool for creating specialized AI agents:

# Create new agent
python .claude/agent-composer.py create my-specialist-agent

# List available agents
python .claude/agent-composer.py list

# View templates
python .claude/agent-composer.py templates

Pre-built Agent Types

• approval-manager - Intelligent approval workflow management
• auto-dev-monitor - System monitoring and health specialist
• satellite-orchestrator - Satellite fleet management
• software-developer - Full-stack development specialist
• data-scientist - Data analysis and ML specialist

🤖 LLM Integration

• Claude: Direct tool calling via MCP protocol
• OpenAI: Function calling integration
• Custom Agents: HTTP REST API endpoints
• WebSocket: Real-time tool execution