Scanr.ai now has an MCP server. Connect Claude Code, Cursor, Codex, VS Code or Claude Desktop, and your agent can turn one plain-English sentence into a scan of every U.S.-listed stock, then look into what it finds.

Connect

You need an API plan, and every plan includes MCP. Create a token with the read scope under Account → API Tokens, then add the server to Claude Code and check it:

claude mcp add --transport http scanr https://mcp.scanr.ai \
  --header "Authorization: Bearer $SCANR_API_TOKEN"

claude mcp list
# scanr: https://mcp.scanr.ai (HTTP) - ✓ Connected

Other clients take the same URL and header in their own config. For Cursor, that's ~/.cursor/mcp.json:

{
  "mcpServers": {
    "scanr": {
      "url": "https://mcp.scanr.ai",
      "headers": {
        "Authorization": "Bearer scanr_..."
      }
    }
  }
}

The MCP Server page has the Codex, VS Code and Claude Desktop configs. Or click Copy onboarding prompt there, paste it into your agent, and it walks you through the setup itself. The claude.ai and ChatGPT custom connectors need an OAuth sign-in, which the server doesn't offer yet.

Ask for a scan

The server gives your agent eight tools: natural_scan, refine_payload, scan, search_securities, market_status, get_news, hot_tickers and ping. In the video, we ask Claude Code:

> Find beaten-down large caps that are starting to turn

The agent calls natural_scan, which turns the sentence into a scan payload and explains its reasoning:

{
  "reasoning": {
    "summary": "Large caps oversold on RSI whose 50-day average has started rising."
  },
  "generated_payload": { … }
}

The payload holds three filters: RSI(14) under 40, a 50-day SMA higher than it was 5 bars ago, and a market cap of at least $10 billion. The agent passes it to scan, which checks every U.S.-listed stock. When we recorded the video, 77 matched in 4.49 seconds, and they're the same 77 you see in the Scanr.ai Charts view. The agent then calls get_news on ABNB and summarises the sentiment of its recent articles.

If you want to change the screen ("only tech", "use the weekly chart"), your agent calls refine_payload, which keeps the filters that were already right. natural_scan and refine_payload spend AI credits. scan doesn't, so running a payload you've already built costs one request.

Tell us what your agent finds at feedback.scanr.ai or [email protected].

— Eoin & The Scanr.ai team