Back to Projects
🚪

Cognigate

Active Open Source Apache 2.0

The open-source enforcement engine for AI agent governance. Receives an action, evaluates policy, returns allow/deny/escalate, and logs a cryptographic proof entry.

View on GitHub API Documentation

Overview

Cognigate is the stateless enforcement engine at the heart of the Vorion ecosystem. It does one thing well: receive an action, evaluate it against policy, return allow/deny/escalate, and log a proof entry. Open source. Embeddable. No opinions — just enforcement.

Named after the Latin "cognitio" (knowledge) + "gate", Cognigate acts as the checkpoint — it reads an agent's trust score and capabilities, enforces the rules, and stamps the receipt.

API Endpoints

POST /evaluate

Evaluate an intent against policy rules. Returns allow/deny decision with reasoning.

GET /trust/:entityId

Get current trust score and level for an entity.

POST /signal

Record a behavioral signal for trust score calculation.

POST /enforce

Enforce a policy decision and record to audit trail.

GET /audit/:entityId

Retrieve audit history for an entity.

Example Request

curl 
curl -X POST https://cognigate.dev/evaluate \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "entityId": "agent_123",
    "intent": {
      "action": "external_api_call",
      "target": "https://api.example.com/data",
      "method": "GET"
    },
    "context": {
      "taskId": "task_456",
      "requestedBy": "user_789"
    }
  }'
Response 
{
  "decision": "ALLOW",
  "trustLevel": 3,
  "trustScore": 612,
  "reasoning": "Agent has Monitored status (T3) with external API capability",
  "conditions": [],
  "proofId": "proof_abc123"
}

Need fleet management and compliance dashboards?

See AgentAnchor →

Related

ATSF Framework BASIS Standard

Vorion Ecosystem: Vorion (standard) → Cognigate (enforcement) → AgentAnchor (governance)