Use cases · for people using AI assistants for research and KYB

Use Indian company data in Claude Desktop and Cursor (MCP)

Last updated 11 November 2026 · by Venu Gopal Reddy Muvva, founder

Add the IndianFirmData MCP server to your AI client and it can search Indian companies, enrich one by CIN, website or name, and check CIN and GSTIN numbers with real, sourced data. For Claude Desktop and Cursor, add a server that runs npx -y github:SurgeEnterpriseAI/indianfirmdata-js with your key in IFD_API_KEY; it forwards every tool call to the hosted server. Clients that support remote MCP can use https://indianfirmdata.com/api/mcp directly. You get 16 tools, at the same credits as the REST API, and the first 3 months are free.

Steps in IndianFirmData

1. Create an API key
Sign up free and create a key in Dashboard → API keys. You need Node.js 18+ installed for npx.
2. Add the server to your client's config
Put the JSON below in claude_desktop_config.json (Claude Desktop) or ~/.cursor/mcp.json (Cursor), with your key in IFD_API_KEY, then restart the app.
3. Ask a question that needs real data
For example: "Find listed pharma companies in Gujarat and give me their CINs" or "Is this GSTIN valid, and which state and PAN does it contain: <GSTIN>?". The assistant calls natural_language_search or decode_identifier and answers from the result.
4. Check what it spent
Every tool call shows on Dashboard → Usage with its credits, at ₹0 during the free period.

Claude Desktop or Cursor config (stdio)

{
  "mcpServers": {
    "indianfirmdata": {
      "command": "npx",
      "args": ["-y", "github:SurgeEnterpriseAI/indianfirmdata-js"],
      "env": { "IFD_API_KEY": "your_key" }
    }
  }
}

Claude Code (remote HTTP)

claude mcp add --transport http indianfirmdata https://indianfirmdata.com/api/mcp --header "Authorization: Bearer $IFD_API_KEY"

The tools your assistant gets

search_companies, natural_language_search, enrich_company, get_company, get_company_people, get_company_jobs, get_company_news, get_company_tech, get_company_tenders, decode_identifier, export_dataset, create_watcher, list_watchers, delete_watcher, list_filters and get_usage. Only the watcher tools change anything; everything else is read-only. Jobs, news and tech are beta, and tenders return an empty list until tender data is available.

Why this beats asking the model directly

Language models know big Indian company names but often get CINs, states, listing details and status wrong, and cannot check a GSTIN's check digit reliably. With the server connected, the answer comes from a tool result that names its source and date, so you can check it.

When IndianFirmData is not the right tool

If your assistant needs people's contact details, live GST status or director (DIN) data, this server does not provide them. And if your client supports neither stdio nor remote MCP, use the REST API from code instead.

Questions

Does it work with ChatGPT?

It works with any client that can connect to an MCP server over stdio or Streamable HTTP. Which ChatGPT plans and modes can add custom MCP servers is up to OpenAI; check its current docs.

Where does the stdio server run?

On your computer, started by your AI client through npx. It holds no data itself: it forwards each tool call to https://indianfirmdata.com/api/mcp with your key, so credits, rate limits and opt-outs are enforced on our server.

Do tool calls cost more than API calls?

No. A tool call costs the same credits as the matching REST call; ID checks are free.

Is the server open source?

Yes, MIT-licensed, in https://github.com/SurgeEnterpriseAI/indianfirmdata-js.

Try it on a company you know.

Start free — 3 months on us