PHANTOM
An autonomous AI command center for penetration testing and security research that runs recursive tool-calling loops against a target with real-time streaming output, a local memory store and a choice of LLM backends.
PHANTOM is an AI assistant built for penetration testing, security research and related general-purpose tasks, run from a phantom CLI command that starts a local server with a browser-based dark interface. Once launched, it lets the connected language model call security tools recursively and continue working toward a goal without needing constant manual prompting.
Tools install their own system dependencies automatically and parse their output for the model, an orchestrator coordinates specialist agents for log analysis, compliance and threat modeling in parallel, and a SQLite-backed memory store gives the agent recall across sessions so it does not repeat reconnaissance. It works with OpenAI, OpenRouter, Ollama, LM Studio, DeepSeek, Claude or any OpenAI-compatible backend, and is installed with a shell or PowerShell script or built manually from source with Node.js.