apurvsinghgautam/robin

What it solves

Robin simplifies dark web Open Source Intelligence (OSINT) investigations by using AI to handle the tedious parts of searching and filtering. It automates the process of finding relevant information across dark web search engines and summarizes the findings into a coherent report, reducing the manual effort required to navigate these environments.

How it works

Robin uses a modular architecture that separates searching, scraping, and LLM processing. It integrates with Tor to access dark web search engines, uses LLMs to refine user queries and filter search results, and provides a summary of the gathered data. Users interact with the tool via a Streamlit-based web interface where they can conduct searches, ask grounded follow-up questions based on the investigation data, or use "one-click pivots" to launch new investigations based on suggested queries.

Who it’s for

It is designed for investigators and researchers conducting lawful OSINT investigations on the dark web.

Highlights

• Multi-Model Support: Compatible with OpenAI, Claude, Gemini, Ollama, and other OpenAI-compatible APIs like Groq or LM Studio.
• Conversational Follow-ups: Allows users to ask questions about a specific investigation without needing to re-run the search.
• One-Click Pivots: Suggests new search queries based on current findings to help investigators pivot their research.
• Docker-Ready: Provides a Docker image for isolated and secure deployment.
• Custom Reporting: Enables saving investigation outputs to files for further analysis.