automagik-tools 0.7.2

A monorepo package for MCP tools with dynamic loading capabilities - featuring Evolution API v2 WhatsApp integration

🪄 AutoMagik Tools

MCP Agents That Learn Your APIs Automagikally

Drop any OpenAPI spec → Get an intelligent MCP agent that learns how you work and gets better with every interaction. Export to code when you're ready to customize.

Born from our daily work at Namastex Labs, AutoMagik Tools creates self-evolving agents that turn any API into a natural language interface.

🧠 Self-Learning MCP Agents

Unlike static tools, AutoMagik agents remember and adapt:

# First time: "How much did I sell last month?"
# Agent learns your sales endpoints, date formats, and preferences

# By the 10th interaction:
# Agent already knows exactly which data you want and how you like it formatted
uvx automagik-tools tool automagik -t sse --port 8000

Three Intelligence Modes:

• 🔧 Standard: Full schema access for precise control
• 📋 Markdown: Agent processes noisy JSON into clean, readable output (powered by GPT-4.1-nano)
• 💬 Genie: Natural language interface - just describe what you need

Powered by GPT-4.1 family models for cost-effective agentic behavior and reliable API reasoning.

🚀 From Any API to Smart Agent

Turn any OpenAPI spec into an intelligent agent:

# Discord API becomes a smart agent
uvx automagik-tools openapi \
  https://raw.githubusercontent.com/discord/discord-api-spec/main/specs/openapi.json \
  -t sse --port 8001

# Share this agent with your entire team via SSE
# Team members can access the same learning agent at http://localhost:8001

Now you can say:

• "Show me unread messages from the design team"
• "Schedule a voice call for 3pm in the dev channel"
• "Find all threads where someone mentioned the new feature"

Agent learns your patterns, server preferences, and communication style.

🧞 Genie: Universal MCP Orchestrator

Genie connects any MCP servers and orchestrates them with persistent memory:

# Run Genie with memory-based agents
uvx automagik-tools tool genie -t sse --port 8000

Configure Genie with Multiple MCP Servers

Genie can orchestrate any combination of MCP servers. Configure via environment variables:

# Method 1: JSON configuration for multiple servers
export GENIE_MCP_CONFIGS='{
  "agent-memory": {
    "url": "http://192.168.112.149:8000/sse",
    "transport": "sse"
  },
  "filesystem": {
    "command": "npx",
    "args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/allowed/directory"],
    "env": {}
  },
  "github": {
    "command": "uvx",
    "args": ["mcp-server-git"],
    "env": {"GITHUB_TOKEN": "your-token"}
  }
}'

# Method 2: AutoMagik-specific shorthand
export GENIE_AUTOMAGIK_API_KEY="your-api-key"
export GENIE_AUTOMAGIK_BASE_URL="http://localhost:8881"
export GENIE_AUTOMAGIK_TIMEOUT="600"

Add Genie to Claude/Cursor (Universal Orchestrator)

{
  "mcpServers": {
    "genie": {
      "command": "uvx",
      "args": [
        "automagik-tools@latest",
        "serve",
        "--tool",
        "genie",
        "--transport",
        "stdio"
      ],
      "env": {
        "OPENAI_API_KEY": "your-openai-key",
        "GENIE_MCP_CONFIGS": "{\"agent-memory\":{\"url\":\"http://192.168.112.149:8000/sse\",\"transport\":\"sse\"},\"filesystem\":{\"command\":\"npx\",\"args\":[\"-y\",\"@modelcontextprotocol/server-filesystem\",\"/allowed/path\"],\"env\":{}}}"
      }
    }
  }
}

Now in Claude/Cursor, Genie can:

• "Use the memory agent to remember this conversation"
• "Check the filesystem for project files and analyze them with the memory agent"
• "Coordinate between multiple tools to complete complex tasks"

🌟 Featured: AutoMagik Orchestration

Enterprise-grade agent orchestration that speaks human:

uvx automagik-tools tool automagik --transport sse --port 8000

Real examples from our users:

• "Monitor inventory across all warehouses and alert me when Corona drops below 10k units"
• "Generate weekly performance reports and send to regional managers"
• "Set up automated quality checks for next month's production runs"
• "Track competitor pricing and notify me of changes above 5%"

💬 Natural Language • 🧠 Memory & Learning • 🔄 Task Orchestration • 🏗️ Framework Agnostic • 👩‍💻 Export to Code

🚀 Quick Start

Environment Setup

Create a .env file with your API keys:

# .env
OPENAI_API_KEY=sk-your-openai-key-here
AUTOMAGIK_API_KEY=your-automagik-api-key
AUTOMAGIK_BASE_URL=http://localhost:8881

# Enable JSON to Markdown processing (optional)
ENABLE_JSON_PROCESSING=true
JSON_PROCESSOR_MODEL=gpt-4.1-nano

Copy this into Claude/Cursor for instant API orchestration:

{
  "mcpServers": {
    "automagik": {
      "command": "uvx",
      "args": [
        "automagik-tools@latest",
        "serve",
        "--tool",
        "automagik",
        "--transport",
        "stdio"
      ],
      "env": {
        "AUTOMAGIK_API_KEY": "YOUR_API_KEY",
        "AUTOMAGIK_BASE_URL": "http://localhost:8881",
        "OPENAI_API_KEY": "YOUR_OPENAI_API_KEY"
      }
    }
  }
}

Where to add:

• Claude Desktop: Settings → Developer → Edit Config
• Cursor: ~/.cursor/mcp.json

Test any API instantly:

# Jira becomes conversational project management
OPENAI_API_KEY=your_key JIRA_API_TOKEN=your_token uvx automagik-tools serve \
  --openapi-url https://dac-static.atlassian.com/cloud/jira/platform/swagger-v3.v3.json \
  --transport sse --port 8002

# Shopify for e-commerce automation  
OPENAI_API_KEY=your_key SHOPIFY_ACCESS_TOKEN=your_token uvx automagik-tools serve \
  --openapi-url https://shopify.dev/docs/api/admin-rest/2023-04/openapi.json \
  --transport sse --port 8003

# SSE mode allows your team to share the same learning agent

📋 Real-World Agent Examples

Genie Orchestrating Multiple Tools

Personal Automation:

• "Use memory agent to remember my GitHub preferences, then check my repos and create a weekly summary"
• "Process these expense receipts with the filesystem tool and store insights in memory"
• "Monitor my crypto portfolio with the trading API and remember my risk preferences"

Team Coordination:

• "Coordinate between Slack and Linear to set up daily standups for the design team"
• "Use the file system to track project deadlines and notify stakeholders via Discord"
• "Analyze customer feedback from multiple sources and create sentiment reports"

Business Intelligence:

• "Compare Q4 sales across all regions using the database and memory tools"
• "Generate inventory reports by coordinating warehouse APIs and document storage"
• "Monitor competitor pricing across platforms and update my preference memory"

The agent learns your patterns - after a few interactions, it knows exactly how you like your data formatted, which metrics matter most, and when to proactively alert you.

🛠️ Advanced Usage

Multiple Server Configurations

Configure Genie with different MCP server combinations:

# Development setup with local tools
export GENIE_MCP_CONFIGS='{
  "filesystem": {
    "command": "npx",
    "args": ["-y", "@modelcontextprotocol/server-filesystem", "/home/user/projects"],
    "env": {}
  },
  "git": {
    "command": "uvx", 
    "args": ["mcp-server-git"],
    "env": {"GIT_AUTHOR_NAME": "Your Name"}
  }
}'

# Production setup with external services
export GENIE_MCP_CONFIGS='{
  "automagik": {
    "command": "uvx",
    "args": ["automagik-tools", "tool", "automagik", "--transport", "stdio"],
    "env": {
      "AUTOMAGIK_API_KEY": "prod-key",
      "AUTOMAGIK_BASE_URL": "https://api.yourcompany.com"
    }
  },
  "slack": {
    "command": "docker",
    "args": ["run", "-i", "--rm", "-e", "SLACK_BOT_TOKEN", "mcp/slack"],
    "env": {"SLACK_BOT_TOKEN": "xoxb-your-token"}
  }
}'

Dynamic MCP Server Connection

You can also pass MCP server configurations directly to Genie via the ask_genie tool:

{
  "tool": "ask_genie",
  "arguments": {
    "query": "List all my GitHub repositories and remember my coding preferences",
    "mcp_servers": {
      "github": {
        "command": "uvx",
        "args": ["mcp-server-git"],
        "env": {"GITHUB_TOKEN": "your-token"}
      },
      "memory": {
        "url": "http://localhost:8000/sse",
        "transport": "sse"
      }
    }
  }
}

🎯 Why Intelligent Agents Matter

The old way: Hours writing API integrations, maintaining complex schemas, fighting with documentation. When APIs change, everything breaks.

The AutoMagik way:

1. Point to any OpenAPI spec
2. Ask in natural language what you need
3. Learn as the agent adapts to your workflow
4. Export to code when you need customization

When your needs evolve, your agent learns and remembers.

🛠️ Built-in Tools

AutoMagik 🤖

AI orchestration that speaks human:

# Quick test with SSE
uvx automagik-tools tool automagik --transport sse --port 8000

What you can do:

• ✨ Use Spark to spawn hives of agents in seconds
• 🔄 Schedule recurring AI tasks and automations
• 💬 Natural language task descriptions
• 🏗️ Works with any AI framework

Genie 🧞

Universal MCP orchestrator with persistent memory:

# Run as SSE server for team sharing
uvx automagik-tools tool genie --transport sse --port 8000

Capabilities:

• 🧠 Persistent memory across all sessions
• 🔗 Connect to unlimited MCP servers
• 💬 Natural language task coordination
• 👥 Shared learning across team members

AutoMagik Workflows 🚀

Smart Claude workflow orchestration with real-time progress tracking:

# Execute Claude Code workflows with progress monitoring
uvx automagik-tools tool automagik-workflows --transport stdio

Features:

• 🚀 Run Claude Code Workflows: Execute workflows with progress reporting
• 📊 Real-time Status: Track workflow completion with visual progress
• 📋 List Workflows: Discover available workflows and recent runs
• 🔄 Status Monitoring: Get detailed workflow execution status

Evolution API (WhatsApp) 📱

Complete WhatsApp automation:

• Send/receive messages
• Media support (images, documents)
• Group management
• Status updates

🚀 The Future: Self-Maintaining Tools

Our roadmap includes agents that:

• Auto-discover API endpoints without OpenAPI specs
• Auto-evolution: Tools that update and adapt when APIs change
• Self-debug and report issues back to maintainers
• Collaborate with other agents for complex workflows
• Generate their own tools from natural language descriptions

Production Deployments (Coming Soon):

• 🐳 Docker: One-click deployment scripts for containerized agents
• ☁️ Cloud Ready: Deploy to AWS, Google Cloud, Azure with pre-built templates
• 🌐 Team Sharing: VPS and network deployments for organization-wide agent access
• 📈 Scalable: Load balancing and auto-scaling for high-demand agents

Coming 2025: A hive of AI agents maintaining this entire codebase - reporting bugs, implementing features, and keeping users happy. All open source.

📋 Real-World Examples

Genie with Memory Agent

{
  "mcpServers": {
    "genie_with_memory": {
      "command": "uvx",
      "args": [
        "automagik-tools@latest",
        "serve",
        "--tool",
        "genie",
        "--transport",
        "stdio"
      ],
      "env": {
        "OPENAI_API_KEY": "your-openai-key",
        "GENIE_MCP_CONFIGS": "{\"agent-memory\":{\"url\":\"http://192.168.112.149:8000/sse\",\"transport\":\"sse\"}}"
      }
    }
  }
}

AutoMagik with Multiple APIs

# Stripe Payments
uvx automagik-tools serve \
  --openapi-url https://raw.githubusercontent.com/stripe/openapi/master/openapi/spec3.json \
  --api-key $STRIPE_API_KEY

# GitHub API  
uvx automagik-tools serve \
  --openapi-url https://raw.githubusercontent.com/github/rest-api-description/main/descriptions/api.github.com/api.github.com.json \
  --api-key $GITHUB_TOKEN

# Your Internal API
uvx automagik-tools serve \
  --openapi-url https://api.yourcompany.com/openapi.json \
  --api-key $YOUR_API_KEY

🛠️ Developer Documentation

Development Setup

# Clone the repo
git clone https://github.com/namastexlabs/automagik-tools
cd automagik-tools

# Install with all dev dependencies
make install

# Run tests
make test

# Create a new tool
make new-tool

Creating Tools from OpenAPI

# Method 1: Dynamic (no files created)
uvx automagik-tools openapi https://api.example.com/openapi.json

# Method 2: Generate persistent tool
uvx automagik-tools create-tool --url https://api.example.com/openapi.json --name my-api
uvx automagik-tools tool my-api

Adding Your Own Tools

1. Create a folder in automagik_tools/tools/your_tool/
2. Add __init__.py with FastMCP server
3. That's it - auto-discovered!

See our Tool Creation Guide for details.

Available Commands

# Core commands
automagik-tools list              # List all available tools
automagik-tools hub               # Serve all tools together
automagik-tools tool <name>       # Serve a specific tool
automagik-tools openapi <url>     # Serve from OpenAPI spec
automagik-tools mcp-config <tool> # Generate MCP config
automagik-tools info <tool>       # Show tool details
automagik-tools version           # Show version

# Development commands  
make install                            # Install dev environment
make test                               # Run all tests
make lint                               # Check code style
make format                             # Auto-format code
make build                              # Build package
make docker-build                       # Build Docker images

🤝 Contributing

We love contributions! See CONTRIBUTING.md for guidelines.

📄 License

MIT License - see LICENSE

🙏 Credits

Built with ❤️ by Namastex Labs

Special thanks to:

• Anthropic for MCP
• FastMCP for the awesome framework
• Our amazing community of contributors

---

Every API becomes a smart agent that learns how you work.
Star us on GitHub • Join our Discord • Follow on Twitter