EAC
Start Building
rocket_launchGetting Started

Embedded Agent Controller

EAC is a policy-enforced action layer between AI and apps. It provides a secure, low-latency bridge that ensures every autonomous decision made by an AI model is validated against your organization's security protocols before execution.

Real-time Policy Enforcement

Validate actions against custom OPA rules in under 15ms.

Audit-Ready Logging

Cryptographically signed logs of every AI interaction and decision.

psychologyCore Concepts

Actions

Primitive operations that the EAC can perform on downstream systems. Each action is typed, versioned, and requires specific permissions.

Engine

The orchestration core that receives model intent, maps it to Actions, and checks the Security layer for authorization.

Security

A multi-tenant policy engine where you define "who" (which model) can do "what" (which action) on "what" (which resource).

sensorsAPI Reference

POST/api/v1/core/engine/process
{
  "intent": "provision_instance",
  "parameters": {
    "provider": "aws",
    "region": "us-east-1",
    "type": "t3.medium"
  },
  "context": {
    "agent_id": "agent_x_04",
    "session_id": "sess_9921"
  }
}

buildImplementation Response

response_payload.json
1{
2"status": "success",
3"trace_id": "tr_01HJ8Z...",
4"policy_check": {
5"result": "ALLOW",
6"eval_ms": 12.4
7},
8"action_output": { ... }
9}