EVA

Why we included this project

EVA is a command-line penetration testing agent that works through an engagement step by step, from initial reconnaissance to exploitation and post-exploitation. The LLM generates the commands and reads the output before deciding what to try next, so rather than getting raw suggestions you get a guided progression with context at every stage. You can point it at a local Ollama model for a self-contained, private assessment, or connect it to OpenAI, Anthropic, Gemini, or a custom endpoint when you want a stronger reasoning model. It keeps persistent sessions, writes reports, and exports attack surface maps, all of which make it practical when you need to document an engagement. Teams doing CTFs, lab practice, or authorized assessments on their own targets will find it a solid assistant, though the project is explicit that EVA guides rather than replaces a human pentester.