What Cabrini Market Data does
Cabrini Market Data MCP server exposes historical US equity market data and SEC records to AI agents. Its coverage includes every US equity from 2003 to the present, with intraday and daily bars, company profiles, quarterly fundamentals, filing indexes, extracted filing sections, Form 4 insider transactions, and joined research briefs.
The service is intended for agents that need to retrieve data during a task rather than rely on a static dataset. Requests are priced individually in USDC on Base through the x402 payment protocol. The documented examples include stock comparisons, earnings-related volume questions, and broader company research.
How it works
MCP clients connect to the hosted endpoint at https://cabrini.ai/mcp. Each paid request can return an HTTP 402 response containing payment instructions. A client signs a USDC transfer authorization, sends it back in an X-PAYMENT header, and receives the requested data after the payment is verified and submitted on Base.
The Python client handles this payment flow when initialized with a Base wallet private key. LangChain and CrewAI helper functions expose the available operations as agent tools. No API key, subscription, or account signup is described as required, but the wallet must hold USDC on Base.
Setup and configuration
For Python applications, install the package with pip install cabrini, then create a Cabrini instance with a private key for a Base wallet. The same private-key argument is used by the LangChain and CrewAI tool helpers.
For MCP use, configure a compatible client with the hosted URL rather than installing a local process. The README shows a remote MCP configuration for Claude and Cursor. Wallet funding can come from bridging to Base, sending USDC from Coinbase, or another supported route; the documentation mentions mainnet USDC and a Base testnet faucet as a separate note.
Tools and capabilities
The documented operations are:
query: full-day intraday bars for one ticker.
daily: daily OHLCV and VWAP at absolute price levels.
batch: one date across multiple tickers.
range: intraday data across multiple trading days.
bars: resampled intraday bars using intervals from 3 to 240 minutes.
scan: screen US stocks using at least one criterion.
tickers: list traded tickers for a date.
company: retrieve a company profile from SEC EDGAR.
fundamentals: retrieve SEC quarterly data.
filings: retrieve SEC filing indexes and, at the higher charge, extracted section text.
insiders: retrieve Form 4 insider transactions.
brief: retrieve a combined research brief.
The Cabrini Market Data MCP server returns intraday values as fractional changes from the session's daily open, not absolute prices. Use daily when absolute open, high, low, close, volume, transactions, or VWAP values are needed. The documented reference includes pre-market activity, so an external 09:30 open may not match the service's daily open.
Limitations and notes
Prices in the endpoint table are examples of the documented charge structure, not fixed quotes. The server returns the live amount requested in each payment response. Intraday methods use percentage changes from the daily open, while daily provides the absolute levels needed to reconstruct prices.
The Cabrini Market Data MCP server requires a funded Base wallet for paid requests. The provided material does not describe free access, offline operation, rate limits, or historical-data adjustments. Screening requires at least one criterion, and the documented bars interval range is 3 to 240 minutes.