I needed an automatic system that could gather and store intelligence with evidence-grade quality.
It's an interesting experiment: using "agentic workflows" for something that actually matters.
Closed-source project.
Damián Búho
Uwindaji Kit
Origin
What it does
Uwindaji can automatically find facts about intelligence targets, keep them tamper-proof, maintain provenance, monitor developments.
You specify a person or organization, you get a report from multiple distributed tools that work together. Every domain, email, phone number and address triggers “specialist” services (non-LLM), which enrich them and share findings between themselves. You can analyze it manually or use it with agents as MCP/CLI (LLMs love to play spies).
Currently this system monitors and prevents sanctions compliance violations.
What's next
- image and audio processing
- more enrichment methods
- HUMINT bridge (chatbot that will process tips and analyze them in context)