Sentinel
Policy intelligence that keeps AI responses on track.
Detect toxicity, unsafe content, NSFW outputs, jailbreak attempts, and policy violations without slowing the rest of your stack down.
Toxic outputs and abuse prevention
Jailbreak and prompt injection detection
Configurable policy scoring and routing
Moderation Request
POST /v1/inspectconst verdict = await slate.sentinel.inspect({
output,
policy: ["toxicity", "jailbreak", "nsfw"],
});
toxicity
0.02
nsfw
0.00
jailbreak
0.93
