AI threat hunting for adversarial prompts and agentic workflows
Nova Hunting is the public entry point for the Nova ecosystem.
Nova is an open source initiative focused on AI security, detection, and threat hunting. It targets adversarial prompts, agent abuse, and misuse patterns across modern AI systems, with an emphasis on visibility and control in agentic workflows.
Website: https://novahunting.ai
The Nova ecosystem brings together tooling, detection logic, and intelligence to support AI threat hunting:
-
Nova Framework
Core engine of the ecosystem. A prompt pattern matching and detection engine designed to identify adversarial behaviors and enforce security policies. -
Nova Docs
Official documentation to install, configure, and start using the Nova framework.
https://docs.novahunting.ai -
Nova Proximity
MCP scanner built on top of Nova, providing inspection and monitoring of agent communications and workflows. -
Nova Rules
Public repository of shared detection rules, enabling community driven research and collaboration.
Nova is connected to PromptIntel, a platform dedicated to cataloging adversarial prompts and maintaining a public classification and taxonomy.
https://promptintel.novahunting.ai
This intelligence layer supports detection research, rule authoring, and continuous improvement of the Nova ecosystem.
Each repository defines its own scope and security disclosure process in its SECURITY.md file.