Phase 1 — Live

Signals

One JSON event per decision. BUY and SELL (including partial sells). Delivered via the Model Context Protocol (MCP) to your AI client. Stable schema, version-pinned.

What you receive

One JSON event per BUY, one per SELL (including partials). Streamed via MCP to your client. Schema is stable.

BUY

{
  "v": 1,
  "event": "buy",
  "ts": "2026-05-18T14:23:11.412Z",
  "ca": "6LiUy31ER...",
  "symbol": "WENGU",
  "entryPriceUsd": 0.00001234,
  "entryMC": 95803,
  "entrySOL": 0.5,
  "discoverySource": "scanner"
}

SELL

{
  "v": 1,
  "event": "sell",
  "ts": "2026-05-18T14:51:09.117Z",
  "ca": "6LiUy31ER...",
  "symbol": "WENGU",
  "exitPriceUsd": 0.00002111,
  "pnlPct": 27,
  "pnlSOL": 0.367,
  "sellPercent": 50,
  "isPartial": true,
  "exitReasonLabel": "Profit-lock target reached"
}

Supported clients

Claude Code

First-class MCP support. Add the server URL + API key to your config and signals stream into your chat.

Setup →

Cursor

MCP server config block in .cursor/config.json. Signals visible in the AI sidebar.

Setup →

Custom client

REST fallback at /v1/mcp/recent returns the last N signals as JSON. Use this if you are not on an MCP-native client.

Setup →

SSE stream

Long-lived Server-Sent Events feed at /v1/mcp/stream for any client that prefers push.

Setup →