Install argusred, log in, and run a read-only Security Scan on a repo you own — three commands, a few minutes.
Install is free. The trailing && argusred (macOS) launches it for the first time, which kicks off sign-up (step 2).
macOS
brew install CosineAI/argusred/argusred && argusred
Linux
curl -fsSL https://raw.githubusercontent.com/CosineAI/argusred-dist/main/install.sh | sh
Windows is coming soon. On Linux, run argusred after install (step 3).
The first run opens a quick Cosine sign-up in your browser — the same login that runs Cosine’s coding agent. New accounts start with 2M free tokens, enough to try a real scan. One account covers both products.
Sign-in is stored locally on your machine, so you won’t be asked again on the next run.
Change into a repo you own and run argusred. A setup screen lets you pick modules, the agent’s permissions, and optional exploit verification; the scan is read-only — it never modifies your code.
When it finishes, the report is written to a single markdown file in your repo:
.argusred/scan-<date>.md
Executive summary, per-module findings, location, severity, cause, and fix direction — for every finding it could ground in your code. The file stays on your machine.