MAA Intelligence

Connect the index to your AI

Let your AI query 418 Korean listed companies and 42,000+ regulatory records directly. No install, no account, no authentication. Just one URL.

Server URL

https://medicalaialliance.org/api/mcp

How to connect

Claude

Claude (app and web)

Settings → Connectors → Add custom connector. Name it 'MAA Intelligence' and paste the URL above. Leave the OAuth fields empty.

After adding it, set the tool permission to 'Always allow' so you are not prompted on every question.

ChatGPT

Settings → Connectors → Add, then paste the same URL. Connector availability depends on your plan.

Cursor

Create .cursor/mcp.json in your project, or ~/.cursor/mcp.json to apply everywhere.

{
  "mcpServers": {
    "maa-intelligence": {
      "url": "https://medicalaialliance.org/api/mcp"
    }
  }
}

VS Code

VS Code (Copilot)

Create .vscode/mcp.json in your workspace.

{
  "servers": {
    "maa-intelligence": {
      "type": "http",
      "url": "https://medicalaialliance.org/api/mcp"
    }
  }
}

CLI

Claude Code and other tools

One command.

claude mcp add --transport http maa-intelligence https://medicalaialliance.org/api/mcp

Any other MCP-capable tool (Windsurf, Zed, Gemini CLI and others) takes the same URL wherever it asks for a remote MCP server or streamable HTTP endpoint.

What you can ask

Which Korean listed companies hold Innovative Medical Device designations?
Show me VUNO's regulatory record and recent financials
Which Korean medical AI companies with FDA clearances have no English website?
How readable are Korean medical AI companies to AI systems?

Four tools

search_companiesFind companies by name, sector or keyword
get_companyOne company in full: overview, products, R&D, financials, regulatory record
search_approvalsSearch MFDS and FDA approvals, trials and designations
industry_overviewSector-wide counts and AI visibility

Notes

Read-only. There are no write operations, and it serves only the public data already published in the index — compiled from KRX, DART, Korea's public data portal and openFDA.

Records are matched to companies by business registration number where the source provides one, otherwise by deterministic name normalisation — never by name similarity. Clinical trial authorisations are pipeline, not approvals, and are counted separately.

Tell us if something is wrong or missing