algothesis-mcp

algothesis-mcp

Research stocks from inside Claude. Real short-interest data, live web search, ticker mapping. No backtests, no fluff.

What it does

• Research any topic — "Kevin Warsh Fed chair", "Trump chip tariffs", "what happened to Nike today" → grounded summary + source links + mapped tickers with live prices
• Real short-interest data — short % of float, days-to-cover, shares short, last report date for up to 20 tickers in one call. Sourced from FINRA via Yahoo (~14-day lag). The data your training-data guesses don't have.
• Find catalysts — upcoming earnings, FDA dates, Fed meetings for any ticker

Install

Add to your Claude Desktop or Claude Code config:

{
  "mcpServers": {
    "algothesis": {
      "command": "npx",
      "args": ["algothesis-mcp"]
    }
  }
}

Restart Claude. Done.

Try it

Just talk to Claude naturally:

"What's the latest on Kevin Warsh and how should I be positioned for a dovish Fed?"

"Show me real short interest on CVNA, BYND, UPST, and GME — which has actual squeeze potential?"

"What stocks could benefit from Trump's chip tariff plans this week?"

"Which biotechs have FDA catalysts this month?"

Claude calls AlgoThesis automatically. No commands to memorize.

Tools

Tool	Use it for
research_topic	Any person, event, macro topic. Live Google Search + ticker mapping with current prices.
get_short_interest	Real squeeze data for up to 20 tickers. Sorted by short % of float.
find_catalysts	Upcoming earnings, FDA dates, Fed meetings.

Why it's useful

Most AI assistants answer market questions from stale training data. AlgoThesis MCP grounds those answers in live web search and current FINRA short-interest data — the two things you can't get from the model alone.

Free to use

Works without an account. Public endpoints, no API key required. Heavy users can create an account at algothesis.ai for the full web app.

Links

• algothesis.ai — Full web app
• npm — Package page

Not financial advice. Short-interest data has typical 10-14 day FINRA reporting lag.