screener-mcp-server
An MCP server for Screener.in to search Indian companies, fetch financial ratios, statements, peer comparisons, and run custom screens.
README
screener-mcp-server
An MCP server for Screener.in — search Indian listed companies, pull key ratios, multi-year financial statements, peer comparisons, and run custom fundamental screens, all from Claude (or any MCP client).
Important: no official API
Screener.in explicitly states they don't provide an API — only CSV export from the UI. This server works by reading the same public, signed-out pages a browser would see (search, company pages, and the /screen/raw/ query endpoint). It does not require login for the core tools.
Because this isn't an official API:
- Please keep request volume reasonable — the server throttles and caches requests for you (5 min cache, ~600ms min gap between requests), don't work around that.
- Guest (logged-out) access to custom screens is capped by Screener to a small number of result rows. If you want more rows or CSV-export-only data, set
SCREENER_SESSION_COOKIE(see below) to a cookie from your own logged-in session. - Screener's HTML structure can change without notice; if a tool starts erroring, the CSS selectors in
src/services/scraper.tslikely need a small update. - Review Screener's Terms before heavy or commercial use.
Tools
| Tool | What it does |
|---|---|
screener_search_companies |
Look up a company's Screener ID/URL by name or ticker |
screener_get_company_overview |
Key ratios (Market Cap, P/E, ROE, ROCE, etc.), About text, Pros/Cons |
screener_get_financial_statement |
Quarterly results, P&L, balance sheet, cash flow, or ratios — multi-year table |
screener_get_peer_comparison |
The peer-comparison table shown on a company's page |
screener_run_custom_screen |
Run a Screener query-builder screen (e.g. Market Capitalization > 500 AND Return on equity > 22) across the whole market |
technical_get_indicator |
RSI, SMA, EMA, or MACD for an NSE/BSE stock, via Alpha Vantage |
Why compute indicators ourselves instead of using Groww or Alpha Vantage?
- Groww's official MCP (
mcp.groww.in) is account-linked via OAuth (it's built to read your portfolio/trades). Folding it into this server would mean building a full OAuth pass-through proxy — a much bigger project, and unnecessary for public technicals like RSI. - Alpha Vantage's NSE/BSE coverage (
NSE:TICKERformat) turned out to be unreliable in practice — it often only returns NASDAQ data despite documentation claiming Indian exchange support.
Instead, technical_get_indicator pulls raw historical close prices from Yahoo Finance's public chart endpoint (ticker + .NS/.BO suffix — the same data source most community NSE tooling, like yfinance, relies on) and computes RSI (Wilder's method), SMA, EMA, and MACD (12/26/9) server-side using standard formulas. No API key, no OAuth, no external rate limits — just one more tool in the same connector.
Since this is an unofficial-but-widely-used endpoint (not a documented Yahoo API), if it ever breaks, check src/services/yahoo.ts — the fix is usually just adjusting the URL or response shape to match Yahoo's current chart endpoint.
Setup
npm install
npm run build
Run locally (stdio) — for Claude Desktop / Claude Code
Add to your MCP client config (e.g. claude_desktop_config.json):
{
"mcpServers": {
"screener": {
"command": "node",
"args": ["/absolute/path/to/screener-mcp-server/dist/index.js"]
}
}
}
Run as a remote HTTP server
TRANSPORT=http PORT=3000 npm start
Then point your MCP client at http://localhost:3000/mcp.
Optional: logged-in session for more screen results
export SCREENER_SESSION_COOKIE="csrftoken=...; sessionid=..."
Grab this from your browser's DevTools → Network tab → any request to screener.in → Cookie request header, while logged into your own Screener account. Never share this value; it's your personal session.
Development
npm run dev # tsc --watch
If Screener changes their page markup and a tool starts failing, check the selectors in src/services/scraper.ts (#top-ratios, .pros/.cons, #profit-loss/#balance-sheet/etc. table sections, #peers table) against the current HTML.
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。