Making LLM security
accessible to everyone.
Sentryɸ exists because pre-deployment security testing shouldn't require a dedicated security team, a six-figure budget, or weeks of manual configuration. We're building the security gate that every AI deployment deserves.
What we believe
Security as a Standard
Every AI model deployed in production should pass adversarial security testing. We believe security scanning should be as routine as unit testing — not a luxury reserved for large enterprises.
Transparency Over Trust
Black-box risk scores create false confidence. Every score in Sentryɸ is explainable: you can trace any finding back to the probe that triggered it, the evidence it produced, and the formula that scored it.
Privacy by Architecture
We don't collect telemetry. We don't require accounts. We don't phone home. Scan data stays on your machine. API keys live in your OS keychain. Period.
Open Source, Always
Sentryɸ is MIT licensed. The scanning engine, scoring algorithm, and deployment advisor are all open source. Security tools should be auditable by the people who use them.
Built in the open
Sentryɸ is open source under the MIT license. Contributions, issues, and security audits are welcome.
View on GitHub →