Megaraptor MCP

A Model Context Protocol (MCP) server that provides AI assistants with access to Velociraptor - the powerful digital forensics and incident response (DFIR) platform.

Overview

Megaraptor MCP enables AI assistants like Claude to interact with Velociraptor servers for:

• Endpoint Management: Search, interrogate, and manage Velociraptor clients

• Artifact Collection: Schedule forensic artifact collection on endpoints

• Threat Hunting: Create and manage hunts across multiple endpoints

• VQL Queries: Execute arbitrary Velociraptor Query Language queries

• Incident Response: Pre-built DFIR workflow prompts for common scenarios

• Deployment Automation: Deploy Velociraptor servers and agents across infrastructure (Docker, binary, cloud, GPO, SSH, WinRM, Ansible)

Features

MCP Tools (33 tools)

Core DFIR Tools (15 tools)

Deployment Tools (18 tools)

MCP Resources

Browse Velociraptor data through standardized URIs:

• velociraptor://clients - Browse connected endpoints

• velociraptor://clients/{client_id} - View specific client details

• velociraptor://hunts - Browse hunt campaigns

• velociraptor://hunts/{hunt_id} - View specific hunt details

• velociraptor://artifacts - Browse available artifacts

• velociraptor://server-info - View server information

• velociraptor://deployments - Browse deployment operations and status

MCP Prompts (8 prompts)

Pre-built DFIR and deployment workflow prompts:

Installation

Prerequisites

• Python 3.10 or higher

• A running Velociraptor server with API access enabled

• API client credentials (see Configuration)

Install from source

git clone https://github.com/yourusername/megaraptor-mcp.git
cd megaraptor-mcp

# Core DFIR functionality only
pip install -e .

# With deployment features
pip install -e ".[deployment]"

# With cloud deployment (AWS/Azure)
pip install -e ".[cloud]"

# All features
pip install -e ".[all]"

Optional Dependencies

Install dependencies manually

# Core only
pip install mcp pyvelociraptor pyyaml grpcio

# For deployment features
pip install paramiko pywinrm cryptography jinja2

# For cloud deployment
pip install boto3 azure-mgmt-compute azure-identity

Configuration

Megaraptor MCP supports two authentication methods:

Option 1: Config File (Recommended)

1. Generate an API client config on your Velociraptor server:

velociraptor --config server.config.yaml config api_client \
    --name mcp-client \
    --role reader,investigator \
    api_client.yaml

2. Set the environment variable:

export VELOCIRAPTOR_CONFIG_PATH=/path/to/api_client.yaml

Option 2: Environment Variables

Set individual configuration values:

export VELOCIRAPTOR_API_URL=https://velociraptor.example.com:8001
export VELOCIRAPTOR_CLIENT_CERT=/path/to/client.crt  # or PEM content
export VELOCIRAPTOR_CLIENT_KEY=/path/to/client.key   # or PEM content
export VELOCIRAPTOR_CA_CERT=/path/to/ca.crt          # or PEM content

API Roles

Assign appropriate roles to your API client based on required capabilities:

Usage

Running the Server

# Using the installed command
megaraptor-mcp

# Or as a Python module
python -m megaraptor_mcp

Claude Desktop Integration

Add to your Claude Desktop configuration (claude_desktop_config.json):

{
  "mcpServers": {
    "velociraptor": {
      "command": "python",
      "args": ["-m", "megaraptor_mcp"],
      "env": {
        "VELOCIRAPTOR_CONFIG_PATH": "/path/to/api_client.yaml"
      }
    }
  }
}

Example Interactions

List connected endpoints:

Use the list_clients tool to show all Windows endpoints

Investigate an endpoint:

Use the investigate_endpoint prompt for client C.1234567890abcdef

Create a threat hunt:

Create a hunt for the file hash a1b2c3d4e5f6... across all endpoints

Run custom VQL:

Run this VQL query: SELECT * FROM pslist() WHERE Name =~ 'suspicious'

VQL Reference

VQL (Velociraptor Query Language) is the core query language. Common patterns:

-- List all clients
SELECT * FROM clients()

-- Search for clients by hostname
SELECT * FROM clients(search='host:workstation')

-- Get running processes from collected data
SELECT * FROM source(client_id='C.xxx', flow_id='F.xxx')

-- Create a hunt
SELECT hunt(artifacts='Windows.System.Pslist', description='Process audit')
FROM scope()

For complete VQL reference, see: https://docs.velociraptor.app/vql_reference/

Deployment Features

Megaraptor MCP includes comprehensive deployment automation for Velociraptor infrastructure.

Server Deployment

Deploy Velociraptor servers using multiple methods:

Example: Deploy Docker server

Deploy a Velociraptor server using Docker on server.example.com with SSH credentials "prod-server"

Agent Deployment

Multiple agent deployment methods for different environments:

Example: Deploy agents via GPO

Generate a GPO deployment package for 500 Windows endpoints using the enterprise profile

Example: Deploy via Ansible

Create an Ansible playbook to deploy Velociraptor agents to all Linux servers in inventory.yml

Deployment Profiles

Pre-configured deployment profiles for different scenarios:

Credential Management

Securely store deployment credentials:

Store SSH credentials for prod-servers with username admin and key file ~/.ssh/prod_key

Credentials are encrypted at rest using AES-256-GCM with a locally-generated key.

Offline Collectors

Build standalone collectors for air-gapped environments:

Build an offline collector for Windows that collects browser history and network connections

Collectors include embedded configuration and can run without network connectivity.

Project Structure

megaraptor-mcp/
├── pyproject.toml           # Project configuration
├── README.md                # This file
├── src/
│   └── megaraptor_mcp/
│       ├── __init__.py      # Package initialization
│       ├── __main__.py      # Module entry point
│       ├── server.py        # MCP server main entry
│       ├── client.py        # Velociraptor API wrapper
│       ├── config.py        # Configuration handling
│       ├── tools/           # MCP tool implementations
│       │   ├── clients.py   # Client management tools
│       │   ├── artifacts.py # Artifact tools
│       │   ├── hunts.py     # Hunt management tools
│       │   ├── flows.py     # Flow/collection tools
│       │   └── vql.py       # VQL query tools
│       ├── resources/       # MCP resource implementations
│       │   └── resources.py
│       ├── prompts/         # MCP prompt implementations
│       │   └── prompts.py
│       └── deployment/      # Deployment automation
│           ├── __init__.py  # Deployment module init
│           ├── tools.py     # Deployment tool implementations
│           ├── server/      # Server deployment
│           │   ├── __init__.py
│           │   ├── binary.py    # Binary deployment
│           │   ├── docker.py    # Docker deployment
│           │   └── cloud.py     # Cloud deployment (AWS/Azure)
│           ├── agent/       # Agent deployment
│           │   ├── __init__.py
│           │   ├── gpo.py       # GPO package generation
│           │   ├── winrm.py     # WinRM deployment
│           │   ├── ssh.py       # SSH deployment
│           │   ├── ansible.py   # Ansible playbook generation
│           │   └── offline.py   # Offline collector builder
│           ├── credentials.py   # Secure credential storage
│           ├── config_generator.py  # Config file generation
│           └── profiles.py  # Deployment profiles (rapid/standard/enterprise)
└── tests/                   # Test suite
    ├── test_config.py
    └── test_deployment.py

Security Considerations

API Security

• API Credentials: Store API client credentials securely. The config file contains private keys.

• Principle of Least Privilege: Use the minimum required roles for API clients.

• Network Security: Ensure API connections are only accessible from trusted networks.

• Audit Logging: Velociraptor logs all API actions. Review logs regularly.

• Quarantine Caution: The quarantine tool can isolate endpoints from the network.

Deployment Security

• Credential Encryption: Deployment credentials are encrypted at rest using AES-256-GCM. The .keyfile is generated locally and should be protected.

• Generated Configs: Server and client configurations contain CA certificates and private keys. These are excluded from git via .gitignore.

• Ansible Playbooks: Generated playbooks may contain CA certificates. Store securely and limit access.

• Cloud Templates: CloudFormation and ARM templates may contain sensitive parameters. Review before committing.

• SSH/WinRM: Use key-based authentication where possible. Avoid storing passwords in plain text.

• Offline Collectors: Built collectors contain embedded configuration. Protect as you would agent binaries.

• GPO Packages: MSI packages contain embedded configuration. Control access to distribution share.

Development

Running Tests

pip install -e ".[dev]"
pytest

Contributing

1. Fork the repository

2. Create a feature branch

3. Make your changes

4. Run tests

5. Submit a pull request

License

MIT License - see LICENSE file for details.

Resources

• Velociraptor Documentation

• Velociraptor GitHub

• VQL Reference

• MCP Specification

• pyvelociraptor

Acknowledgments

• The Velociraptor team at Velocidex for creating an amazing DFIR platform

• Anthropic for the Model Context Protocol specification