# OpenOSINT

An AI-powered OSINT agent for authorized security research, with twenty investigation tools behind a natural-language interface and an MCP server so any compatible AI client can drive them.

OpenOSINT is an open-source intelligence agent for security researchers and analysts, built so a natural-language request calls one of twenty investigation tools rather than the model guessing at results: the AI issues a tool call and the actual binary runs, so a hallucinated finding is not possible. It can be used as an interactive REPL, a direct command-line tool with no AI involved, a Model Context Protocol server, or a browser web interface.

It installs with pip and works with Claude, GPT-4, OpenRouter or local Ollama models supplied by the user. A separate hosted option runs individual OpenOSINT tools, such as email, username and domain reconnaissance, as pay-per-use cloud actions with no local install, and that usage funds the self-hosted project's development. It is written in Python, licensed under MIT, and is intended for authorized security research only.

Website: https://openosint.tech