I built an MCP-native OSINT framework that lets AI agents investigate from your terminal

You give Claude a single prompt — "investigate this email address" — and it autonomously chains five tools: email enumeration, username search across 300+ platforms, breach lookup, WHOIS, and IP geolocation. No manual invocations, no copy-pasting output between scripts, no babysitting. That's what OpenOSINT enables, and it works because the entire tool surface is exposed through the Model Context Protocol.

What is OpenOSINT?

OpenOSINT is a Python framework that acts as an MCP server, exposing 9 OSINT tools to any MCP-compatible AI client — Claude Code, Claude Desktop, or anything else that speaks the protocol. It is not a scraper, not a dashboard, not a GUI. It is a structured tool surface that lets LLMs call real intelligence-gathering utilities the same way they call any other function.

The framework wraps proven OSINT tools — holehe, sherlock, sublist3r, phoneinfoga, HaveIBeenPwned — in async, stateless Python functions that the MCP layer can discover and invoke.

Version 2.1.0 is live on PyPI, the repo is MIT-licensed, and it runs on Python 3.10+. If you know what OSINT is but have never had an AI agent drive an investigation autonomously, this is the fastest path to that experience.