Developer Integration
Deploy the Reality Engine across your agentic fleet in minutes.
Endpoint
https://realitykernel.dev/v1/check
Auth Type
Bearer Token
1. Authentication Protocol
All requests must be authenticated via the Authorization header. Use your SOC-issued API key as a Bearer token.
Authorization: Bearer YOUR_RK_API_KEY
2. Causal Verification Request
Before any agent executes a system-level command, it must clear the Reality Engine. The engine simulates the command across multiple causal trajectories.
Request Parameters
| Field | Type | Description |
|---|---|---|
command | string | The raw system command or query to be verified. |
prime_intent | string | The agent's stated objective for this action. |
session_id | string | (Optional) Unique identifier for the agent session. |
agent_id | string | (Optional) Unique identifier for the specific autonomous agent instance. |
3. Reference Implementation (Python)
Integrate the verification step into your agent's execution loop.
import requests
def verify_action(command, intent):
url = "https://realitykernel.dev/v1/check"
headers = {"Authorization": "Bearer rk_live_..."}
payload = {
"command": command,
"prime_intent": intent
}
# Initialize causal simulation
response = requests.post(url, json=payload, headers=headers)
data = response.json()
# Handle reflexive collapse
if data['verdict'] == 'BLOCK':
print(f"Reflexive Collapse Triggered: {data['evidence']}")
return False
return True
4. Verdict Semantics
The engine returns a deterministic verdict based on the simulation's basin mapping.
ALLOW
Safe trajectory. Command verified as benign or read-only.
WARN
Suspicious trajectory. Simulation shows potential divergence. Human review recommended.
BLOCK
Reflexive collapse. Action crossed the separatrix into a destructive state. Execution halted.
