Shibui Finance is a free MCP server that connects Claude to 64 years of US stock market data. It covers 9,500+ NYSE and NASDAQ companies with 31M+ daily price records (1962 to present), quarterly and annual financial statements, daily valuations, and 56 pre-calculated technical indicators. You connect it to Claude once and then ask questions in plain English. Claude queries the database directly and returns structured results. No downloads, no scripts, no API keys. Setup takes under 2 minutes on both Free and Pro plans.
The old way: CSV uploads and API scripts
If you have tried to use stock data in Claude before, you probably followed one of these paths. Download a CSV from Yahoo Finance, clean up the columns, upload it to Claude, and hope the file is not too large. Or write a script to pull prices from a data provider, export to CSV, and upload that. Or build something more elaborate with an API wrapper so Claude can call your local server.
These approaches work. People use them every day. But they share the same limitations: you are restricted to whatever data you thought to download in advance, each upload is a one-time snapshot that goes stale immediately, file size limits constrain how much history you can include, and the setup requires at least some technical skill. Comparing two companies means two downloads. Screening across 9,500 companies means a very large file or giving up entirely.
The fundamental problem is that Claude has no persistent connection to the data. Every session starts from scratch. You are the middleware - downloading, formatting, uploading, and repeating whenever you need a different slice.
The MCP approach: connect once, ask anything
MCP (Model Context Protocol) is Anthropic's standard for connecting Claude to external data sources. Instead of uploading files, you give Claude a persistent connection to a database or service. Claude can then query that source directly, on demand, as part of any conversation. Shibui Finance is an MCP server purpose-built for financial data.
"What was Apple's revenue growth rate each quarter for the last 3 years?"
Claude queries the Shibui database directly and returns the results in a structured format. No download step, no script, no file. The data is always current (updated daily, end-of-day), and you can follow up with any question about any company without uploading anything new.
What data is available
- Daily prices - OHLCV data from 1962 to present, 31M+ records across 9,500+ securities
- Quarterly and annual financial statements - income statement, balance sheet, and cash flow from 1990 to present
- Daily valuations - market cap, P/E, P/B, enterprise value from 1993 to present
- 56 pre-calculated technical indicators - RSI, MACD, SMA, Bollinger Bands, and more, computed daily for every stock
- Quarterly earnings - actual vs. estimate EPS, surprise percentage, and report dates
- ~9,500 US-listed securities - stocks, ETFs, ADRs, and REITs on NYSE and NASDAQ
See the full data documentation for column-level detail on all 12 tables.
Three things you can do that CSV uploads cannot
The difference between a file upload and a database connection becomes obvious when your questions span the full market.
"Show me the 10 largest companies by market cap in the Technology sector with their current P/E ratios"
This query runs across the entire universe of 9,500+ companies, joining valuation data with sector classification, in under a second. With CSV uploads, you would need to have downloaded valuation data and sector data for every company in advance.
"Find companies where operating margin exceeded 20% every quarter for 3 years"
This checks 12 individual quarters per company across thousands of companies. That is tens of thousands of data points evaluated in a single query. No CSV file you could practically upload would contain all the quarterly financials needed.
"What was the average stock price change 30 days after an earnings miss of more than 10%?"
Event studies like this join earnings data with daily price data at specific time offsets across thousands of historical events. The query scans millions of price records to compute the answer. This kind of cross-table, cross-market analysis is effectively impossible with uploaded files.
Limitations to know about
Shibui covers US equities only. There is no international market data, no options, no crypto, and no index-level data.
All data is end-of-day. There are no real-time quotes and no intraday prices. If you need live prices during market hours, use TradingView or your broker's platform.
The underlying data comes from tier-3 data providers, not Bloomberg or S&P Capital IQ. Coverage and accuracy are good for most analytical work, but this is not institutional-grade data. Extreme values (returns above 1000%) sometimes reflect data errors or corporate actions rather than real moves.
Query results are capped at 200 rows. For broader screens, you need to aggregate or narrow your filters. For international markets, Koyfin or TIKR are better options.
Works with Claude Free and Pro
MCP connections work on every Claude plan, including the free tier. Shibui Finance itself is free. You connect once through Claude's settings and the connection persists across conversations. The setup instructions walk you through it in under 2 minutes.