XFlux MCP server

Use the Model Context Protocol to give AI agents access to the XFlux read API — search, profiles, and timelines. Monitors and webhooks are still configured in the Dashboard; the MCP layer is intentionally thin.

What this is (and isn't)

Is: on-demand tweet/user lookups for coding agents and research workflows.
Isn't: a replacement for account monitors, streaming, or webhook delivery — use monitors for always-on alerts.

Available tools

xflux_get_user — profile by @username
xflux_search_tweets — search recent posts
xflux_get_user_tweets — user timeline
xflux_get_tweet — single tweet by ID

Install & run

Source lives in the repo at mcp-server/. You need Node 18+ and an API key from the Dashboard.

cd mcp-server
npm install
export XFLUX_API_KEY=xflux_your_key_here
node index.mjs

Cursor configuration

Add to Cursor MCP settings (Settings → MCP → Edit config). Use the absolute path to index.mjs on your machine:

{
  "mcpServers": {
    "xflux": {
      "command": "node",
      "args": ["/path/to/xflux/mcp-server/index.mjs"],
      "env": {
        "XFLUX_API_KEY": "xflux_your_key_here"
      }
    }
  }
}

Claude Desktop

Same JSON block in Claude Desktop config (claude_desktop_config.json). Restart the app after saving.

Monitors + MCP together

Typical pattern: use MCP during development to explore accounts and craft keyword filters, then set up trading keyword templates in the Dashboard for production alerts. Route webhooks to Make.com per the Make guide.

API limits

MCP calls count against your plan quota like any REST request. See Plans & Limits.