Security work,
grounded in evidence.
Audit code, investigate signals, and validate detections with a security toolkit built for your AI coding agent.
Open source • Offline-first • Agent Skills compatible
$ tarosec simulate suspicious-login
Scenario auth.suspicious-login
Mode synthetic
Events generated 5
DETECTION OUTCOME
Inconclusive
No independent rule observations supplied.
Evidence and report saved locally.
The toolkit
One entry point.
Six ways to work.
Start with $tarosec. Choose a specialist when you know what you need.
A practical approach
Follow the evidence.
Keep the context.
Tarosec helps your agent document what it saw, what it can conclude, and what still needs verification.
Review supplied source, records, and telemetry with read-only workflows.
Separate observations from interpretation and mark missing evidence clearly.
Compare synthetic scenarios with recorded detections and retain evidence references.
Install Tarosec
GitHub release v0.2.0 includes seven skills, including Report.
Run these commands from your project directory. You need Node.js for installation and Python 3.10+ to run the included helpers.
Start a new session in your coding agent and enter $tarosec ./project to begin.
npx skills add refactorroom/tarosec
npm install tarosec@0.1.0
The npm release is currently v0.1.0 and downloads six skill directories to node_modules/tarosec/skills/.
npx skills@1.9.0 experimental_sync --copy --agent codex
If you chose npm, run this next to copy the skills from your installed dependencies into this project's Codex skill directory. The command uses Skills CLI version 1.9.0.
GitHub