Agent Trust Levels (NATL)
4 minLast updated: 2026-03-31
What Is NATL?
NATL (Neimar Agent Trust Levels) is a zero-trust security framework where every agent has a trust level that controls what actions it can take. Higher trust = more autonomy.
The 5 Trust Levels
- L0: OBSERVE: Read and analyze only. Cannot respond to users or modify data. Used for monitoring agents.
- L1: RESPOND: Can reply to users and use RAG knowledge base. Cannot create records or call external APIs.
- L2: ACT: Can create records, publish portal events, and modify data within its scope.
- L3: ORCHESTRATE: Can use tools, call webhooks, trigger other agents. Full workflow capabilities.
- L4: AUTONOMOUS: Full autonomy including scheduled tasks, unsupervised operations, and cross-system actions.
Default Trust Levels by Agent Type
- L0: Analyst, Portal (observe only)
- L1: General (chat), Document, Website, Voice (respond to users)
- L2: Security (can block threats)
- L3: Orchestrator, Workflow (can use tools and trigger actions)
Agents can be promoted above their default level after meeting requirements.
Plan Limits on Trust
- Free / Starter: Maximum L1: agents can only respond
- Pro: Maximum L3: full workflow and orchestration
- Max / Enterprise: Maximum L4: fully autonomous agents
Trust API
GET /api/v1/agents/:id/trust - Get trust config and stats
PUT /api/v1/agents/:id/trust - Update trust configuration
GET /api/v1/agents/:id/trust/history : View trust level change history
GET /api/v1/agents/:id/trust/incidents: View security incidents