llm_osint

A proof-of-concept for using an LLM plus web search to perform OSINT — it researches public figures and builds profiles on them autonomously, then performs a task with that information. By sshh12. It is the single-LLM pole of this spoke's AI-OSINT pattern: one model in a research loop, the simplest form of the idea that Kallisto-OSINTer scales up to multi-agent orchestration. T3 at ingest, superseded 2026-08-20 → T1: a first-party PoC repo.

What it is

The loop is the whole idea: an LLM is given a target, queries the web, reads results, and assembles a profile — automating the manual reconnaissance an analyst would otherwise do by hand with OSINT tooling. As a proof-of-concept it demonstrates the pattern (LLM + tools → autonomous intelligence-gathering) rather than shipping a polished product.

Why it matters here

llm_osint is the founding seed of this spoke — the first source parked in the hub _inbox (2026-06-21) that flagged "LLM-powered OSINT / AI-security" as a cluster to watch. With kallisto-osinter (multi-agent) and awesome-social-engineering (the human-element resource base), it completed the 3-source cluster the spoke spun out from.

Ethics note

Autonomous profiling of real people from public data is the spoke's central dual-use tension — useful for authorized investigation/defense, hazardous as surveillance. Catalogued, not endorsed (see synthesis).

Related

ai-osint · kallisto-osinter · osint · social-engineering · llm-deanonymization-at-scale · synthesis

Re-tiered 2026-08-20 → T1. Under the corpus rule adopted this day (../QUALITY.md → tier by the claim, not the URL), this page rests on what the proof of concept does — one model in a research loop — which is the whole artifact. A project is the primary source for its own design, and the earlier grade tiered the URL rather than the claim.