Open-Source Intelligence
OSINT & Intel Dashboards
Collection and analysis tooling that turns scattered public data into a single monitored picture.
The problem
OSINT tooling is fragmented and most of it leaks your queries to a third party. I wanted collection and analysis that runs locally, with the reasoning layer on a local model instead of a hosted API.
How it's built
- Collectors for news, market, and geopolitical feeds normalized into a single event stream
- Geospatial layer mapping real-world infrastructure from OpenStreetMap data
- Recon automation for subdomain enumeration and asset discovery
- Local LLM (Ollama) analysis layer — zero API cost and nothing leaves the box
- Agent skills that chain collection, scoring, and dossier generation
Outcome
A private intelligence workflow where collection, enrichment, and reasoning all run on hardware I control.
At a glance
- Local LLM
- Reasoning layer, zero API cost
- 0 bytes
- Query data sent to third parties
- Name → dossier
- Automated collection to scored report
Decisions & tradeoffs
Considered: A hosted model for analysis
Chose: Ollama running locally
Why: OSINT queries are sensitive; sending them to an API leaks the investigation.
Tradeoff: Smaller models and local GPU/CPU limits on throughput.
Considered: Separate tools per collection source
Chose: Normalizing everything into one event stream
Why: Correlation is the whole value; separate tools cannot correlate.
Tradeoff: An ingestion schema to maintain as sources change shape.