StocksDeepDive

AI / MCP

Esta página está en inglés porque documenta un servidor MCP técnico de solo lectura; StocksDeepDive también existe en español - mira Deep Dive o la API en español.

StocksDeepDive runs an MCP (Model Context Protocol) server - the same read-only, public stock-score data as the JSON API, exposed as callable tools for AI assistants like Claude and ChatGPT instead of raw HTTP endpoints. No API key, no sign-in, nothing here is user data.

Endpoint: https://stocksdeepdive.com/mcp (Streamable HTTP transport).

Tools

  • deep_dive(ticker) - computed value/quality/psychology/discovery/moat scores for one stock.
  • compare(tickers) - the same scores for up to 10 stocks side by side.
  • scan(universe, top_n) - the ranked overnight scan for a whole index (e.g. “ASX 200”, “S&P 500”), highest score first. Most universes update nightly; a few large ones update weekly instead - see https://stocksdeepdive.com/api’s “Coverage & update cadence” section for which is which.
  • moat(ticker) - the cached Moat Score and erosion flag for one stock.
  • research_notes(ticker) - any StocksDeepDive research notes written about that company, with links.

Every tool result includes attribution, disclaimer and as_of/link fields, same as the JSON API - please attribute and link back to the site when you surface this data to someone.

Connecting a client

Most MCP clients (Claude Desktop, Claude Code, and others that support Streamable HTTP) accept a server URL directly. Point yours at:

https://stocksdeepdive.com/mcp

No headers, no authentication, no configuration beyond the URL.

Prefer just reading pages?

https://stocksdeepdive.com/llms.txt is a plain-Markdown index of the site’s worthwhile pages, meant to be fetched directly instead of crawling the whole site - a lighter option than either the API or MCP when all an assistant needs is to read, not call tools.

Terms

Free for any use, commercial or not, with attribution and a link back to stocksdeepdive.com. Nothing served here is financial advice - see the disclaimer in every tool result. No user data (portfolios, watchlists, emails) is ever served here or ever will be.

Míralo funcionar en una empresa real

Ingresa un ticker en Deep Dive, compara dos lado a lado, o lee la investigación hecha a mano en Rational Compounder. Lo nuevo se publica en el blog (en inglés).