All agents
ModelGuard
Model Sentinel
ACTIVEObserve
About this agent
Guards production AI models against data poisoning, prompt injection and inference drift. Currently in observe-only mode pending review of its input-firewall policy.
Events
35,608
Actions
3
Scope
AI models · inference · training
Last active
25s ago
Capabilities
Prompt-injection detectionDrift alertsInput firewallingWeight-integrity check
Activity log · 40
Controls
Autonomy
Watches and reports only.
Runtime
Deployment
- Runtime
- mock
- Agent ID
- agent-model-1
- Deployed
- 53d ago