ShrikeDrive the governance layer yourself.
Pick a scenario and watch Shrike decide, action by action. The scans are real: the same 9-layer pipeline that handles live customer traffic. Detection internals stay private by design.
Live shared demo. Scans here leave an audit trail: the same per-action evidence Shrike keeps for governed agents. Use sample data, not production secrets.
GCP Marketplace · approved9-layer cascade130+ detector typesOWASP LLM Top 10 mappedThe taxonomy behind the verdicts.
~130 internal detector types reduced to 16 categories, rolled up into their OWASP LLM Top 10 buckets. Click any category for its detector types and attack-class examples.
About the taxonomy
Every scan verdict names one of these categories, never a raw detector: mechanism stays private, meaning stays legible. The rollup is what maps a verdict to your compliance evidence: category → OWASP bucket → control.
Click any category on the board for its detector types and attack-class examples.
Powered by Shrike's action governance layer · Docs · Trust Center · Get started free · Deployment options
Detection internals (9-layer cascade, per-layer attribution, pattern logic) stay private by design. Attack classes and category descriptions are described, not reproduced, and are drawn from the OWASP LLM Top 10 (2026) and prior public security research. Sandbox verdicts demonstrate the live pipeline; they are not an assessment of your systems.