A self-OSINT footprint scanner named after the hundred-eyed giant. Run it on yourself, see what the internet can see about you, and fix it before someone else finds it first.
PythonOSINTGemini 2.0 FlashSearch groundingHosted web appMCP server
Argus started as a single command-line tool. It grew into a small family, so whichever way you like to work, the same consent-first scan is right there. Same engine underneath, three ways in.
Shipped Argus MCP, the scanner wrapped as a Model Context Protocol server. Now any agent that speaks MCP can run a footprint scan as a first-class tool, no shelling out. It closes the loop on the family idea: the same engine is a CLI for me, a web app for you, and a tool for the machines. Lives at github.com/lachydotmcg/argus-mcp.
Supargus goes live
The hosted web version is up at supargus.netlify.app. Same consent-first scan, none of the terminal. It is the version you can actually hand to a non-technical friend and say "go look yourself up", which was always the point of a self-OSINT tool.
Two-pass OSINT engine
The scan runs in two passes. First, local async checks fan out across 28 sites to enumerate where a username exists. Then a second pass hands the picture to Gemini 2.0 Flash with Google Search grounding for live, cited web intelligence, so the report points at real sources instead of guessing. It all lands in a standalone HTML/JSON report you keep.
Consent first, by design
OSINT tooling is a double-edged thing, so the design choice came before the code: Argus is built to be run on yourself. A consent gate sits in front of every scan, the framing throughout is "find your own exposure before someone else does", and the defensive use case is the whole product, not a disclaimer bolted on the end. This is the part that is really mine.