mcp-stock-server (FinanceExpert)

A small Model Context Protocol server built with FastMCP. It exposes Yahoo Finance market data (via yfinance) so assistants can fetch quotes, history as CSV, and plain-text Unicode price charts.

Requirements

• Python 3.11+

• uv (recommended) or another way to install dependencies from pyproject.toml

Setup

cd mcp-stock-server
uv sync

Run (stdio)

The server speaks MCP over stdio (the default for mcp.run()):

uv run python server.py

Add as an MCP server (Cursor)

1. Open Cursor Settings → MCP (or edit your MCP JSON config — often ~/.cursor/mcp.json on macOS/Linux).

2. Register a stdio server whose working directory is this repo and whose command starts server.py.

Option A — uv run (recommended)

Replace /absolute/path/to/mcp-stock-server with the real path where you cloned the repo.

{
  "mcpServers": {
    "FinanceExpert": {
      "command": "uv",
      "args": ["run", "python", "server.py"],
      "cwd": "/absolute/path/to/mcp-stock-server"
    }
  }
}

After uv sync, this uses the locked dependencies from uv.lock.

Option B — project virtualenv Python

Use this if you prefer not to invoke uv from the MCP client:

{
  "mcpServers": {
    "FinanceExpert": {
      "command": "/absolute/path/to/mcp-stock-server/.venv/bin/python",
      "args": ["server.py"],
      "cwd": "/absolute/path/to/mcp-stock-server"
    }
  }
}

Create the venv first (from the repo root): uv sync (installs deps into .venv).

3. Save the config and restart Cursor (or reload MCP). In the MCP panel you should see FinanceExpert with tools get_stock_analysis, get_historical_prices, and get_stock_price_chart.

Other MCP clients (e.g. Claude Code, editors with MCP support) use the same idea: command + args + cwd for a stdio server.

Tools

get_stock_price_chart parameters

• ticker — Symbol, e.g. INTU, AAPL.

• days — Used when period is omitted: Yahoo range Nd (calendar days). Default 30.

• period — Optional Yahoo period string; when set, it overrides days. Examples: 10y, 5y, 1y, 6mo, ytd, max.

Examples:

• Last month of sessions (by calendar days): days=30

• Last ten calendar years (Yahoo window): period="10y"

Market data comes from Yahoo through yfinance; semantics match yfinance history(period=...).

Disclaimer

Quotes and history are informational only, not investment advice. Yahoo data can lag or contain errors; verify independently for decisions.