NOVA MCP lets you configure NOVA (the prompt pattern matching) as a guardrail for your AI system.

What is NOVA MCP?

NOVA MCP is a prompt-validation service designed to enhance the security of AI systems by acting as a guardrail. It checks incoming prompts against predefined Nova rules to block any that match high-severity security or jailbreak patterns.

How to use NOVA MCP?

To use NOVA MCP, clone the repository, install the dependencies, and configure your LLM credentials in a .env file. Run the server and use the provided client script to send prompts for validation.

Key features of NOVA MCP?

• Rule-based filtering of prompts using NOVA rules.
• Configurable rules directory for custom rules.
• Structured JSON logging of authorized and blocked prompts.
• Compatibility with various LLM clients including OpenAI and Anthropic.

Use cases of NOVA MCP?

1. Ensuring safe interactions with AI by blocking harmful prompts.
2. Customizing security rules for different applications.
3. Logging and monitoring prompt validation for compliance and auditing.

FAQ from NOVA MCP?

• Can NOVA MCP be used with any LLM?

Yes! NOVA MCP supports multiple LLM clients including OpenAI, Anthropic, and Azure OpenAI.

• How do I add new rules?

You can add or edit .nov files in the nova_rules/ directory and restart the server to load changes.

• Is NOVA MCP open source?

Yes! NOVA MCP is released under the MIT License and is available on GitHub.