Choose the capability layer.

The capability layer gives partners a controlled way to embed AI Security LLC where they need it, without adopting the full interface. They can take a single bounded capability or compose a small, governed module set that fits partner-native workflows while preserving clear authority boundaries, retrieval limits, and replayable evidence. The plane supports AI-native code paths, multi-step attack reasoning, adversarial validation, structured findings, qualified attack paths, and remediation state, all backed by the AI Security LLC OEM engine, authority graph, and evidence replay.