Skip to content

API Reference

Jump to bottom
cvsz edited this page Sep 1, 2026 · 1 revision

📡 API Reference

Base URL: http://localhost:8080

Authentication: set ZASI_API_KEY env var; send X-API-Key: <key> header (public endpoints exempt).

Rate limit: 60 requests/minute per IP on all /api/* routes (429 Too Many Requests if exceeded).

GET Endpoints

Endpoint Auth Description
GET / No React SPA cockpit (index.html)
GET /jarvis No SPA route — J.A.R.V.I.S. chat page
GET /subsystems No SPA route — 176-subsystem catalog
GET /cockpit No SPA route — quantum hardware runner
GET /mcp No SPA route — MCP JSON-RPC console
GET /ws Yes WebSocket upgrade — 2s telemetry push
GET /api/status No System status (public, always 200)
GET /api/telemetry Yes CPU, RAM, GPU, Arc Reactor, global phi
GET /api/tick Yes Execute one autonomous cognitive cycle
GET /api/subsystems Yes Full 176-subsystem catalog
GET /api/openapi.json No OpenAPI 3.0 specification
GET /api/webhooks Yes List registered event webhooks
GET /api/execute/<key> Yes Execute named subsystem by key

POST Endpoints

Endpoint Body Description
POST /api/jarvis/chat {message, persona} Multi-turn chat (JARVIS/FRIDAY/EDITH)
POST /api/jarvis/stream {message, persona} SSE word-by-word streaming response
POST /api/mcp JSON-RPC 2.0 MCP protocol dispatcher
POST /api/mutate {variable, delta} Hot-mutate cognitive state variable
POST /api/rsi/upgrade {version} Deploy 320× RSI hot-swap upgrade
POST /api/webhooks {url, event} Register webhook (event: tick

WebSocket Push Format

Every 2 seconds all connected /ws clients receive:

{
  "type": "telemetry",
  "timestamp": 1234567890.123,
  "cpu_load": 12.4,
  "memory_used_mb": 3210,
  "arc_reactor_gw": 178.2,
  "state": {"x": 20, "y": 30},
  "logs": [...]
}

MCP JSON-RPC Methods

Method Description
tools/list List all available MCP tools
tools/call Execute a named tool with params
resources/list List MCP resources

Clone this wiki locally