S&P Global MCP Server

Official S&P Global MCP server (built by Kensho) for natural-language access to S&P Capital IQ Financials, market data, transcripts, and company intelligence.

Finance by S&P Global / Kensho OAuth2 active

Connect (Remote Server) GitHub Docs

Overview

The S&P Global MCP server, built and operated by Kensho (an S&P Global company), exposes the LLM-Ready API over the Model Context Protocol so that AI agents can query S&P Global datasets in natural language. It covers S&P Capital IQ Financials, Market Data, Business Relationships, Earnings Call Transcripts, Company Intelligence, and M&A Transactions, with company lookup by ticker, ISIN, or CUSIP and access to over 200 financial line items including revenue, EBITDA, and R&D expense.

The server can be consumed two ways. The recommended path is the managed remote endpoint at https://kfinance.kensho.com/integrations/mcp, which is pre-integrated with Claude, ChatGPT, Microsoft Copilot Studio, Amazon Quick Suite, Databricks, and Mistral (Beta). For teams that need full control, the open-source kensho-kfinance Python package ships a local MCP server (python -m kfinance.mcp) that wraps the same LLM-Ready API.

Access requires an active S&P Global Market Intelligence subscription or trial. The server supports three auth modes (refresh token, OAuth client credentials with a key pair, and an interactive browser flow) and standard MCP transports (stdio, SSE, streamable HTTP). This is the canonical, vendor-built integration: any other "S&P Global MCP" repos in the wild are community projects, not official.

Tools

Tool	Description
`Company identification`	Resolve and retrieve company profiles via ticker, ISIN, or CUSIP, including industry classification, employee count, HQ location, and operating status.
`Market data / prices`	Fetch daily, weekly, monthly, or yearly OHLC prices, volumes, market cap, and enterprise value with adjustments.
`Financial statements`	Pull complete balance sheets, income statements, and cash flow statements across multiple periods.
`Financial line items`	Extract specific line items from 200+ metrics such as revenue, EBITDA, R&D expense, and derived ratios.
`Segments`	Return revenue and operating income breakdowns by business unit and geography.
`Business relationships`	Identify suppliers, customers, and partnerships extracted from SEC filings.
`Earnings`	Access earnings call transcripts and earnings event data.
`Estimates`	Retrieve analyst estimates and consensus figures.
`Capitalizations`	Return capitalization structure data for a company.
`Competitors`	List peers and competitors for a given company.
`Mergers and acquisitions`	Query M&A transactions involving a company.
`Rounds of funding`	Return private market funding round data.
`Key developments`	Surface curated company key developments and events.
`CUSIP and ISIN lookup`	Map between security identifiers (CUSIP, ISIN, ticker).

Setup Guide

Prerequisites

An S&P Global Market Intelligence subscription or trial. Contact S&P Global to request access to the LLM-Ready API.
Credentials issued by Kensho (refresh token, or client ID + private key for OAuth client credentials).

Option A: Remote (managed) server

Point your MCP client at the hosted endpoint:

https://kfinance.kensho.com/integrations/mcp

In Claude (org admin):

Go to Admin settings, Connectors, Add custom connector.
Enter the URL above, name it (for example "S&P Global"), and confirm.
Individual users can then enable the connector and authenticate.

Option B: Local server via `kensho-kfinance`

Requires Python 3.10+.

pip install kensho-kfinance

Run the server (default transport is SSE; use --stdio for Claude Desktop):

python -m kfinance.mcp --stdio

Claude Desktop config (claude_desktop_config.json):

{
  "mcpServers": {
    "kfinance": {
      "command": "<path_to_python>",
      "args": [
        "-m",
        "kfinance.mcp",
        "--stdio"
      ]
    }
  }
}

Authentication flags (local server)

Refresh token (dev, expires after 7 days): --refresh-token <token> (obtain from kfinance.kensho.com/manual_login)
Key pair (recommended for production): --client-id <id> --private-key <key>
Browser OAuth (default fallback): omit auth args and a browser flow will launch

Transports

--stdio / -s: stdio (Claude Desktop, MCP Inspector)
--sse: Server-Sent Events (default)
--streamable-http: HTTP transport

Use Cases

Pull a side-by-side income statement and cash flow comparison for a peer group of public companies for a competitor benchmarking memo.
Track margin and EBITDA trends across multiple quarters for a target company before an investment committee.
Build a supply chain risk view by enumerating disclosed suppliers and customers of a portfolio company from SEC filings.
Summarize the last earnings call transcript for a ticker and extract management commentary on guidance.
Screen M&A activity or funding rounds in a specific sector over a date range.

Example Prompts

"Pull the last 5 years of revenue, EBITDA, and free cash flow for AAPL, MSFT, and GOOGL and show them in a table."
"Show me daily OHLC prices for NVDA for the last 90 days, adjusted for splits."
"Who are the disclosed suppliers and customers of Tesla according to recent SEC filings?"
"Summarize the most recent earnings call transcript for Snowflake and list any guidance changes."
"List all M&A transactions in the cybersecurity sector where the target had revenue over $100M in the last 12 months."

Pros

Official integration built by Kensho, an S&P Global company, so data is sourced directly from S&P Capital IQ.
Broad coverage: financials, market data, transcripts, business relationships, M&A, estimates, and key developments through one server.
Multiple deployment modes (managed remote endpoint or self-hosted Python server) and multiple auth flows (refresh token, client credentials, browser OAuth).
Pre-built connector integrations for Claude, ChatGPT, Copilot Studio, Amazon Quick Suite, Databricks, and Mistral.

Limitations

Requires a paid S&P Global Market Intelligence subscription; you must contact S&P Global to provision credentials, there is no free tier.
Tool definitions are generated dynamically from the kFinance client's LangChain tools, so the canonical tool list is not enumerated in the README and may shift between versions.
Refresh tokens expire after 7 days, so production deployments effectively require setting up the OAuth client credentials flow with a private key.

Alternatives

FMP MCP server (community), backed by Financial Modeling Prep for a lower-cost financials API.
Polygon.io MCP server for real-time and historical market data.
Bloomberg or FactSet via custom MCP wrappers if you already license those terminals; no official MCP server is published by either vendor as of this writing.