Warning Migration from v1.x (Python): The Python MCP server previously published on PyPI as
investor-agentis deprecated. v2.0 is a complete rewrite in TypeScript on Cloudflare Workers. The PyPI package will not receive further updates.
A financial research MCP server that exposes a single codemode tool. Instead of calling 13 separate tools, the LLM writes JavaScript code that orchestrates multiple data-fetching functions in a single call -- fetching prices, calculating indicators, and combining results in one round trip.
{
"mcpServers": {
"investor-agent": {
"type": "url",
"url": "https://investor.ferdousbhai.com/mcp"
}
}
}- Runtime: Cloudflare Workers with Durable Objects for persistent MCP sessions
- Sandbox: @cloudflare/codemode v0.2.0
DynamicWorkerExecutorruns LLM-generated JavaScript in a sandboxed V8 isolate - Caching: Cloudflare KV with per-function TTLs (1 minute to 24 hours)
- Data sources: yahoo-finance2 v3, CNN Fear & Greed, alternative.me Crypto Fear & Greed, NASDAQ API
- Technical analysis: trading-signals (SMA, EMA, RSI, MACD, Bollinger Bands)
The server exposes a single MCP tool called codemode. It accepts a code parameter containing JavaScript that runs in a sandboxed V8 isolate. All 13 functions are available on the codemode object and accept a single argument object.
| Function | Description |
|---|---|
fetchJson({ url, headers? }) |
Fetch JSON from a URL with retry and browser headers |
fetchText({ url, headers? }) |
Fetch raw text/HTML from a URL with retry and browser headers |
toCleanCsv({ rows }) |
Convert an array of objects to CSV, auto-removing empty columns |
validateTicker({ ticker }) |
Validate and normalize a ticker symbol (trims, uppercases) |
periodToDates({ period }) |
Convert a period string (1d, 5d, 1mo, 3mo, 6mo, 1y, 2y, 5y, 10y, ytd, max) to { period1, period2 } dates |
quoteSummary({ symbol, modules }) |
Fetch Yahoo Finance quote summary (supports 27 modules: assetProfile, financialData, price, recommendationTrend, earningsHistory, etc.) |
getHistorical({ symbol, period1, period2?, interval? }) |
Fetch historical OHLCV price data. Interval: 1d, 1wk, 1mo |
getOptions({ symbol, date? }) |
Fetch options chain. Omit date to get available expirations |
getMarketMovers({ category?, count?, session? }) |
Top gainers, losers, or most active stocks |
getCnnFearGreed({}) |
CNN Fear & Greed Index with sub-indicator scores |
getCryptoFearGreed({}) |
Crypto Fear & Greed Index from alternative.me |
getNasdaqEarnings({ date?, limit? }) |
NASDAQ earnings calendar for a given date |
calculateIndicator({ ticker, indicator, period?, timeperiod?, ... }) |
Calculate SMA, EMA, RSI, MACD, or BBANDS with configurable parameters |
The code parameter must be an async arrow function expression — the runtime wraps it as (CODE)().
Fetch the current price of AAPL and calculate its 14-day RSI in a single call:
async () => {
const [summary, rsi] = await Promise.all([
codemode.quoteSummary({ symbol: "AAPL", modules: ["price"] }),
codemode.calculateIndicator({ ticker: "AAPL", indicator: "RSI", timeperiod: 14, numResults: 1 })
]);
const price = summary.price.regularMarketPrice;
const latestRsi = rsi.values[rsi.values.length - 1];
return { ticker: "AAPL", price, rsi: latestRsi.rsi, date: latestRsi.date };
}Compare market and crypto sentiment:
async () => {
const [cnn, crypto] = await Promise.all([
codemode.getCnnFearGreed({}),
codemode.getCryptoFearGreed({})
]);
return { cnn: cnn.fear_and_greed, crypto };
}pnpm install
# Create KV namespace (first time only)
wrangler kv namespace create CACHE
# Copy the returned id into wrangler.jsonc under kv_namespaces
# Local development
pnpm run dev
# Deploy
pnpm run deploy
# Testing
pnpm run test
pnpm run test:watch
pnpm run typecheckMIT