Finance Toolkit MCP Server is a project by Jeroen Bouma that connects large language models to a structured financial analysis engine. Rather than letting a model answer stock questions from training memory, it exposes a defined set of tools the assistant can call, so return on invested capital, valuation multiples and technical indicators are calculated on demand.
The server bundles more than 500 metrics into 22 tools, with three additional tools dedicated to search. Coverage spans valuation models, technical indicators and macroeconomic data, and the same request produces the same number whether it is issued from Claude, ChatGPT, Copilot or Cursor. A startup router pattern and a setup wizard handle configuration, while OAuth and API key support cover authentication.
It suits analysts, finance students and developers who already use an AI assistant for research and want reproducible figures instead of plausible-sounding estimates. Anyone comparing companies, testing an investment case or pulling macro series can query in natural language and receive computed output, as the sample ASML, European banks and US economic cycle conversations illustrate.
You can use the free hosted server or run it locally, and the MIT license permits modification. Start with the installation section, connect your preferred client, then begin with a narrow question such as a single valuation ratio before moving to multi-metric comparisons. The FAQ and Under the Hood pages explain routing and mod
Comments & Ratings
0No comments yet. Be the first to review.