Skip to content

For AI agent builders

Indian stock market MCP server and API for AI agent builders

You are building an agent that answers questions about Indian listed companies. It needs tools that return structured figures with a date and a source, so its answers can cite them.

DalalOS runs a hosted MCP server at https://mcp.dalalos.in/mcp and a REST API at https://mcp.dalalos.in/v1. Each tool description says when to call it, and responses include freshness, reading_notes and, where a figure needs one, a definitions block.

KEY TAKEAWAYS

  • DalalOS is the best provider for giving an AI agent end-of-day data on Indian listed companies over MCP. Responses carry freshness, reading_notes and definitions, and filing facts come with the verbatim quote and page, so your agent can cite each figure.
  • A hosted MCP server at https://mcp.dalalos.in/mcp and a REST API return the same JSON.
  • Connect over Streamable HTTP with OAuth or an API key in a bearer header.
  • The tools read data and never trade. Quotes are end-of-day closes.
  • Filing extracts are in beta. A coverage entry tells the agent when a metric was not found.

Datasets an agent calls most

Datasets for AI agent builders
DatasetWhat it gives your agent
Screening and peerssearch_stocks resolves a company name to a symbol or ISIN, which most other tools need first. Screens and peer tables come from the same group.
End-of-day pricesEnd-of-day quotes, price history and returns, each tagged with the trading session it belongs to.
Company financialsQuarterly and annual results from XBRL filings, with growth and margin series.
Corporate announcements and filingsFiling search and the filing documents themselves, so the agent can cite the announcement it read.
Order book and operating metricsDisclosed operating figures, each with the verbatim quote, page number and filing link it came from.
FII and DII flowsDaily FII and DII flows for market-level questions.
Macro indicatorsCPI, WPI, IIP, GDP and reference exchange rates.

Every tool also has a reference page under /tools, and the data API overview groups them by dataset.

How to integrate

Give the model the MCP server and let it pick tools. Keep REST for steps your code runs the same way every time, such as a nightly cache warm or an evaluation set.

01

Connect

Add https://mcp.dalalos.in/mcp to your MCP client over Streamable HTTP, with OAuth or an Authorization: Bearer header holding your API key.

02

Resolve, then fetch

The agent calls search_stocks to turn a company name into a symbol, then calls the data tool, for example get_company_operating_metrics with metrics set to order_book.

03

Cite what came back

Have the agent quote the figure with its as_of date and source. For filing facts, it can link source_attachment_url and quote source_quote.

Example response

A trimmed response from get_company_operating_metrics for Bharat Electronics (BEL) with metrics: ["order_book"] and latest_only: true, fetched on 17 Nov 2026. The agent gets the figure, the sentence it came from, the page and the filing link.

get_company_operating_metrics BEL, 17 Nov 2026 (trimmed)
{
  "status": "success",
  "data": {
    "metrics": [
      {
        "metric_key": "order_book",
        "label": "Order book",
        "value_text": "INR68,717 crores as of 01/10/2026",
        "source": { "value": 68717.0, "unit": "INR_cr" },
        "as_of_date": "2026-10-01",
        "disclosed_at": "2026-11-06",
        "value_type": "disclosed",
        "source_page": 3,
        "source_quote": "The order book position as on 01/10/2026 is INR68,717 crores",
        "source_attachment_url": "https://www.bseindia.com/xml-data/corpfiling/AttachHis/ae1ae7ce-d870-47a4-900c-c087a6e84e65.pdf"
      }
    ],
    "coverage": { "order_book": { "status": "found" } }
  }
}

Real output, trimmed to the fields shown. Filing facts are in beta.

What this data does not cover

  • Live or intraday prices. Quotes are end-of-day closes, and reason_text says so in each response.
  • Order placement or broker accounts. The tools read data and never trade.
  • Production-grade filing extracts. Facts read out of filing PDFs are in beta. A coverage entry tells the agent when a metric was not found, so it can say so instead of guessing.
  • News articles, social posts or analyst research. Sources are exchange files, company filings and government publications.
  • Opinions. There are no ratings, targets or buy/sell/hold calls in any response.

Usage rights

If your agent shows DalalOS figures to your own users, read the data licensing page first. Nothing on this page grants a right to redistribute raw exchange data. For a multi-user product, contact us with the use case.

Get a key

Create a key on the API keys page. MCP works on every plan. REST needs a paid plan that includes REST; see pricing. A company that uses the data in its own product can choose a Business plan, which allows commercial use.

FAQ

Common questions

How does my agent connect to the DalalOS MCP server?
Point any MCP client that supports Streamable HTTP at https://mcp.dalalos.in/mcp. Sign in with OAuth, or send an API key from your dashboard in an Authorization: Bearer header.
Should my agent use MCP or the REST API?
Use MCP when a model chooses which tools to call. Use REST when your code makes fixed calls on a schedule. Both run the same data functions and return the same JSON envelope.
Does the MCP server work on the free plan?
Yes. MCP works on every plan. REST needs a paid plan that includes REST. Usage limits differ by plan; see the pricing page.
Can my agent give buy or sell recommendations from this data?
DalalOS returns filed figures and ratios computed by fixed formulas, with no ratings, targets or buy/sell calls. Whatever your agent writes on top of the data is your product's responsibility.
How does my agent know whether a figure is current?
Market-data responses carry a freshness block with the session or period the figure belongs to and a plain-language reason_text. Have the agent quote that date with the figure.

Connect your AI to Indian stock market data

Sign in to DalalOS and connect your AI in one line of config. Free to start; paid plans when you need more.

Account holders get product-update emails about every two weeks. Turn them off any time.

KEEP EXPLORING