OPFOR
OPFOR is an open-source red-teaming tool for AI agents, LLM applications and MCP servers, generating attacks across the OWASP LLM, Agentic AI, MCP and API Security top-ten lists and judging each response with an LLM, from a command line, an IDE, or a browser extension.
OPFOR is named for the military term for the unit that plays the attacker in training exercises. It red-teams the whole AI agent surface, prompts, tools, MCP servers, memory and multi-turn reasoning, generating targeted attacks for the OWASP LLM Top 10, OWASP Agentic AI Top 10, OWASP MCP Top 10, OWASP API Security Top 10, and EU AI Act bias test suites, then firing them at a target and judging each response with an LLM.
It can be run one-shot from the command line, or set up once with a saved configuration reusable in continuous integration; a browser extension lets a non-developer red-team a deployed chatbot with no code or environment variables, and it can also run as an MCP server so a coding agent in Cursor or Claude Desktop can red-team another agent through natural language. Every attack prompt, request, response and judge verdict is logged, so a run is reproducible, auditable and forkable.
It is written in TypeScript, released under the Apache license, and requires an API key from a supported model provider to run.