What is the MCP Firewall?

The MCP Firewall is a Model Control Protocol (MCP) compatible service available on the UBOS Asset Marketplace that provides a text filtering firewall with a powerful rules engine to protect data and enforce policies when using Large Language Models (LLMs) like Claude.

How does the MCP Firewall work?

The MCP Firewall acts as a filtering layer between your applications and LLMs. It processes text data against a customizable rules engine before it reaches the LLM, allowing you to identify patterns, enforce policies, and transform text content.

What are the key features of the MCP Firewall?

The key features include a powerful rules engine with pattern matching, default rules for sensitive information, customizable replacements and transformations, a REST API for integration, MCP protocol support, persistent storage using SQLite, enhanced Docker support, and Smithery compatibility.

How do I get started with the MCP Firewall using Docker?

To get started with Docker, clone the repository, build the Docker image, and run the container with persistent storage using the provided commands or the deploy_docker.sh script. Detailed instructions are provided in the documentation.

How do I integrate the MCP Firewall with Claude?

To integrate with Claude, configure the .mcp.json file to point to the MCP Firewall server, specifying the URL, transport, timeout, protocol version, and tools.

Can I add custom rules to the MCP Firewall?

Yes, you can add custom firewall rules via the API. The MCP Firewall supports both regex and plain text pattern matching, allowing you to define specific patterns and replacements.

What is the UBOS platform, and how does it relate to the MCP Firewall?

UBOS is a full-stack AI Agent Development Platform that helps you orchestrate AI Agents, connect them with your enterprise data, build custom AI Agents with your LLM model, and create Multi-Agent Systems. The MCP Firewall, available on the UBOS Asset Marketplace, enhances the security of AI Agents built on the UBOS platform.

What kind of API Endpoints the MCP Firewall has?

The MCP Firewall has API endpoints for text processing like /process and /redact, rules management like /rules (GET, POST, PUT, DELETE, RESET) and system health with /health and /.

What are the security considerations when using the MCP Firewall?

Key security considerations include understanding the performance trade-offs between regex and plain text matching, ensuring the quality and comprehensiveness of your rules, respecting text size limits, validating rules, and integrating the MCP Firewall as part of a broader security strategy.