You know the OSINT workflow. Open a terminal. Run holehe against an email. Copy a username you found. Switch tools. Run sherlock . Open a browser. Check HaveIBeenPwned manually. Pull up a WHOIS tab. Take notes. Repeat. Every tool is a silo. Every pivot is manual. The investigation logic lives entirely in your head. I wanted to fix that. What I built OpenOSINT is an open-source Python framework with an AI agent at its core. You describe a target in natural language — an email address, a username, a domain, an IP, a phone number — and the agent decides which tools to run, chains them based on what it finds, executes everything against the real binaries, and compiles a structured Markdown report. Three interfaces: Interactive AI REPL (default) — type natural language, agent chains the tools autonomously Direct CLI — run individual tools directly, no AI, perfect for scripting MCP Server — expose all 9 tools to Claude Code or Claude Desktop The demo Here's a real session. No mocking.…