Fastify + Swagger — 40+ endpoints with API key auth, rate limiting, and a WebSocket push channel.
Docs are English-only for v0.1.0. Spanish and French translations of the full reference land in v0.2. The marketing site already supports EN · ES · FR — only this docs zone is EN-first.
REST API exposing every Vizzor capability programmatically. Full Swagger documentation at /docs when the server is running.
1 vizzor api start --port 3100 # Start server 2 vizzor api key create "my-app" # Create API key All endpoints require an X-API-Key header. The server is rate-limited to 300 req/min per key .
1 curl -H "X-API-Key: $VIZZOR_API_KEY" \ 2 https://api.vizzor.ai/v1/market/price/BTC Generate keys with vizzor api key create <name>; revoke with vizzor api key revoke <name>. Keys are stored in the SQLite api_keys table — no plaintext leakage to disk.
GET /health # Health check (public) GET /docs # Swagger UI POST /v1/chat # AI chat (SSE streaming) POST /v1/chat/thread # Threaded chat reply GET /v1/conversations # List conversations GET /v1/conversations/:id # Get conversation with messages GET /v1/market/price/:symbol # Price GET /v1/market/prices?symbols= # Batch prices GET /v1/market/trending # Trending tokens GET /v1/market/fear-greed # Fear & Greed Index GET /v1/market/ml-health # ChronoVisor engine status GET /v1/market/trenches # Trenches scanner GET /v1/chronovisor/:symbol # ChronoVisor prediction GET /v1/chronovisor/predictions # List all predictions GET /v1/chronovisor/accuracy # Prediction accuracy metrics GET /v1/polymarket/opportunities/:symbol # Edge opportunities for symbol GET /v1/polymarket/positions # Open orders + recent trades POST /v1/polymarket/trade # Place trade on market GET /v1/notifications # List notifications GET /v1/notifications/:id # Get notification detail POST /v1/notifications/mark-read # Mark all as read DELETE /v1/notifications/:id # Delete notification GET /v1/notifications/rules # List alert rules POST /v1/notifications/rules # Create alert rule DELETE /v1/notifications/rules/:id # Delete alert rule POST /scan # Token security scan POST /trends # Market trends POST /track # Wallet forensics POST /predict # AI prediction POST /audit # Contract audit POST /v1/backtest # Backtest GET /v1/agents # List agents POST /v1/agents # Create agent POST /v1/agents/:name/start # Start agent POST /v1/agents/:name/stop # Stop agent GET /v1/agents/:name/wallet # Agent wallet balance GET /v1/agents/:name/trades # Agent trade history GET /v1/agents/:name/stats # Agent trade statistics GET /v1/portfolio/:id # Portfolio POST /v1/agents/emergency-stop # Global kill switch WS /ws # WebSocket real-time push POST /v1/chat returns Server-Sent Events. Pass conversationId to continue an existing thread; omit it to create a new conversation:
1 curl -N -H "X-API-Key: $VIZZOR_API_KEY" \ 2 -H "Content-Type: application/json" \ 3 -d '{"message":"predict BTC 4h","conversationId":"abc-123"}' \ 4 https://api.vizzor.ai/v1/chat The stream emits event: token frames during generation and event: tool_call frames when the AI invokes a tool (price fetch, scan, prediction, alert-arm). A final event: done frame carries the full canonical response.
WS /ws upgrades to a real-time stream. The server pushes:
New prediction emissions (per tier)
Prediction resolutions (HIT / MISS / NEUTRAL)
Price-alert hits
Agent decisions + trades
All payloads use the same canonical shape as the REST responses — clients can dedupe by id.
Every error returns:
{ "error" : { "code" : "rate_limited" , "message" : "Rate limit exceeded — 300 req/min" , "retryAfter" : 42 } } Codes are stable; messages are human-readable. retryAfter is set on rate_limited and service_unavailable.
