yahoo-finance-mcp-server
An MCP server that exposes comprehensive Yahoo Finance data (quotes, financials, options, news, and more) to AI agents via the yfinance library, requiring no API key.
README
Yahoo Finance MCP Server
An MCP server that exposes the full Yahoo
Finance data surface to AI agents, backed by the
yfinance library. No API key
required.
Note on the "Yahoo Finance API": Yahoo retired its official public finance API years ago. This server uses
yfinance, which wraps Yahoo's internalquery1/query2.finance.yahoo.comendpoints (handling cookie/crumb auth for you). It is the most complete free way to access Yahoo's data, but it is unofficial — endpoints can change or rate-limit without notice.
What the agent can see
Every major data category Yahoo Finance serves is exposed as a tool:
| Tool | Data |
|---|---|
get_quote |
Fast price snapshot (last/open/high/low, volume, market cap, 52-wk range) |
get_ticker_info |
Full company/security profile (.info): sector, ratios, margins, targets, summary |
get_isin |
ISIN identifier |
get_historical_prices |
OHLCV history by period or date range, any interval (1m → 3mo) |
download_multiple |
Batch historical prices for many symbols at once |
get_corporate_actions |
Dividends, splits, capital gains |
get_financial_statements |
Income statement / balance sheet / cash flow (annual & quarterly) |
get_earnings |
Earnings calendar & dates, EPS/revenue estimates, EPS trend & growth |
get_analyst_data |
Recommendations, price targets, upgrades/downgrades |
get_holders |
Major / institutional / mutual-fund holders + insider activity |
get_shares |
Historical shares outstanding |
get_sustainability |
ESG risk scores |
get_option_expirations |
Available option expiry dates |
get_option_chain |
Calls & puts (strike, IV, OI, volume, bid/ask) for an expiry |
get_news |
Recent related news articles |
get_sec_filings |
Recent SEC filings with document links |
get_fund_data |
ETF/mutual-fund holdings, allocations, sector weights, bond ratings |
search |
Resolve a name → symbol; matching quotes + news |
lookup |
Enumerate instruments by type (stock/etf/index/currency/crypto/…) |
get_sector |
Sector overview, top companies/ETFs, industries |
get_industry |
Industry overview, top performing/growth companies |
get_market_status |
Market open/closed status & index summary by region |
screen_predefined |
Preset screens (day_gainers, most_actives, …) |
screen_custom |
Custom numeric screen (e.g. market cap > $1B) |
Tickers use Yahoo symbols: AAPL, MSFT, BTC-USD, ^GSPC, EURUSD=X.
Installation
Requires Python 3.10+.
python3 -m venv .venv
source .venv/bin/activate
pip install -e .
Running
The server speaks MCP over stdio:
yahoo-finance-mcp
# or
python -m yahoo_finance_mcp
Claude Desktop / Claude Code config
Add to your MCP client config (e.g. claude_desktop_config.json):
{
"mcpServers": {
"yahoo-finance": {
"command": "/absolute/path/to/yahoo-finance-mcp-server/.venv/bin/python",
"args": ["-m", "yahoo_finance_mcp"]
}
}
}
Or, with Claude Code:
claude mcp add yahoo-finance -- /absolute/path/to/.venv/bin/python -m yahoo_finance_mcp
Deployment (HTTP, port 8081)
For remote/agent access the server can run over streamable HTTP instead of stdio. Transport and binding are controlled by env vars:
| Var | Default | Purpose |
|---|---|---|
MCP_TRANSPORT |
stdio |
streamable-http (or sse) to serve over HTTP |
MCP_HOST |
0.0.0.0 |
bind address |
MCP_PORT |
8081 |
listen port |
Run it directly:
MCP_TRANSPORT=streamable-http MCP_PORT=8081 python -m yahoo_finance_mcp
# endpoint: http://<host>:8081/mcp
Automated deploy via SSH
deploy/deploy.sh ships the code to a remote host, installs it, and (re)starts
it on port 8081 with a health check. It authenticates over SSH using these
GitHub repository secrets:
| Secret | Meaning |
|---|---|
DEPLOY_SSH_HOST |
remote endpoint (host/IP) |
DEPLOY_SSH_PORT |
SSH port |
DEPLOY_SSH_USER |
SSH username |
DEPLOY_SSH_PASSWORD |
SSH password |
Python is auto-provisioned. The script needs Python ≥ 3.10. It first scans
the remote for any installed interpreter that qualifies (python3.10–3.13,
including /usr/local/bin). If none is found — e.g. on Raspberry Pi OS
Bullseye, which ships only Python 3.9.2 — it installs build deps and compiles
CPython (default 3.11.9, override with PYTHON_BUILD_VERSION) to
/usr/local via make altinstall. This one-time bootstrap takes ~15–40 min on
a Pi; subsequent deploys detect the installed interpreter and skip it.
Building requires root: it works if the deploy user has passwordless sudo,
otherwise it falls back to sudo -S using DEPLOY_SSH_PASSWORD. You can still
set PYTHON_BIN to force a specific interpreter; it's used if it qualifies.
The .github/workflows/deploy.yml workflow runs the script on every push to
main (and on manual workflow_dispatch). To run it by hand:
DEPLOY_SSH_HOST=1.2.3.4 DEPLOY_SSH_PORT=22 \
DEPLOY_SSH_USER=pi DEPLOY_SSH_PASSWORD=secret \
PYTHON_BIN=python3.11 \
./deploy/deploy.sh
The script clones/updates the repo, builds a venv, stops any prior instance on
the port, starts the server detached (setsid, survives disconnect, logs to
server.log), then verifies POST /mcp returns 200.
Notes & limitations
- Unofficial data source. Yahoo may rate-limit or change responses. Tools
return a structured
{"error": ...}payload instead of crashing when a call fails, so the agent can react. - Intraday history is range-limited by Yahoo (e.g.
1mdata only for the last few days). - Data is provided for informational purposes; respect Yahoo's Terms of Service. Not investment advice.
Development
pip install -e .
python -c "from yahoo_finance_mcp.server import mcp; print(len(mcp._tool_manager.list_tools()), 'tools')"
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。