VulnClaw

AI-driven penetration-testing CLI that takes a natural-language objective and autonomously runs the full workflow of information gathering, vulnerability discovery, exploitation and report generation. Uses a goal-driven solver over a blackboard graph state model that only accepts verified tool output as evidence, and drives an MCP toolchain (fetch, browser and Burp) across web and CTF-oriented skills. Supports many OpenAI-compatible LLM providers and offers CLI, TUI and web interfaces.