Frequently Asked Questions
Common questions about Sentinel and AI safety.
General
What is Sentinel?
Sentinel is a Decision Firewall for AI Agents — a safety layer that validates
every input and output of your AI systems using the THSP protocol (Truth, Harm,
Scope, Purpose).
How does Sentinel differ from content moderation?
Content moderation focuses on filtering inappropriate content. Sentinel goes further
by implementing a complete safety framework that considers:
Does Sentinel slow down my AI application?
The heuristic layer adds <10ms latency, which is negligible for most applications.
The optional semantic layer adds 1-5 seconds but provides deeper analysis for
edge cases. You can configure which layer to use based on your needs.
Technical
What AI frameworks does Sentinel support?
Sentinel integrates with:
See our Integrations page for the full list.
Can I customize the safety rules?
Yes! Sentinel supports:
How do I handle false positives?
1. Use sentinel.validate(..., strict=False) for softer validation
2. Add specific patterns to your allowlist
3. Report the issue — we continuously improve our detection
Pricing & Support
Is Sentinel free to use?
Sentinel offers:
How do I get support?
Security
Is my data sent to external servers?
You can disable the semantic layer for fully local operation.
Does Sentinel store my data?
No. Sentinel processes data in-memory and does not persist:
Enterprise customers can enable audit logging with encryption.
How do I report a security vulnerability?
Please email security@sentinelseed.dev with:
We respond within 24 hours and follow responsible disclosure practices.
Governance
What is $SENTINEL token?
$SENTINEL is the governance token for the Sentinel ecosystem. Token holders can:
How do I participate in governance?
1. Hold at least 100,000 $SENTINEL tokens
2. Connect your wallet at /governance
3. Vote on active proposals
4. Create proposals (requires 5M tokens)
See our Governance page for more details.