Skip to content

kkfir13/warren-buffett-brain

v1.3.0MIT

Open-source, source-linked Warren Buffett and Berkshire Hathaway investment research for ChatGPT: real decisions, investing principles, historical mistakes and Buffett-inspired stock analysis.

Warren Buffett Brain

Research businesses with a Buffett-inspired framework. Check the evidence, question the moat, and separate a good company from a good price.

Warren Buffett Brain is a free, independent, open-source library of historical Berkshire Hathaway research and reusable instructions for AI assistants. It is not Warren Buffett's actual brain, an official Berkshire product, a live-stock-price feed, or an AI that knows what Buffett would buy today.

Start with ChatGPT · Browse the research website · Read the eight case studies · Instructions for AI agents

Try it in ChatGPT right now

You can copy this prompt into ChatGPT today. No Warren Buffett Brain installation or account is needed to try the methodology:

Analyze [COMPANY / TICKER] using Buffett-inspired value-investing principles.

Explain how the business earns money, the evidence for and against its moat, management's capital allocation, owner-like cash flow, debt and the strongest bear case. Separate the quality of the business from whether the current share price offers a margin of safety.

Use dated primary filings and show your sources. Separate reported facts, your interpretation and assumptions. Do not invent quotes, prices or fair values, and do not pretend Buffett endorses this stock. Finish with the three most important things to investigate next.

Important: this is a normal ChatGPT prompt, not an installed skill. Pasting it does not load this repository's historical notes. Ask your assistant to read the linked research files only if it actually has access to the repository or you provide the files.

Choose how to use the project

What you wantAvailable now?Start here
Ask ChatGPT to analyze a company using the frameworkYes — ordinary prompt, no installationCopy the prompt
Browse Buffett/Berkshire historical evidence and original case studiesYes — public repository and websiteResearch index · Case studies
Have an AI agent use this repository's actual filesYes, if that agent can read these filesAI agent guide
Import the historical research as a ChatGPT Personal SkillEligible accounts/workspaces onlyImport instructions
Connect a local MCP server from a desktop clientYes, with Python and a compatible stdio MCP clientLocal MCP setup
Install Warren Buffett Brain directly from the public ChatGPT Plugins DirectoryNot yet — the plugin is not publicly listedAvailability explained

The prebuilt Skills-only plugin ZIP is a package for plugin distribution, not a one-click public ChatGPT installation. The separate Personal Skill ZIP is for users whose ChatGPT workspace supports skill uploads. They are different archive formats; neither includes a hosted MCP connection.

What's in the research library?

  • 55 independently written notes about documented investing principles, Berkshire decisions, mistakes and historical context.
  • 60 external source links, including the Berkshire shareholder-letter index. A link does not mean a whole letter, meeting recording or book was copied or reviewed.
  • Eight original long-form case studies, including GEICO, See's Candies, Coca-Cola, Goldman Sachs, BNSF and Berkshire's mistakes.
  • Source-linked research records with evidence, interpretation, limitations and a trail back to original publishers.
  • An offline Python search script and optional local MCP tools: search, fetch, and research_overview.

Try questions like: What did See's Candies reveal about pricing power? / What went wrong with Dexter Shoe? / Does GOOGL generate enough cash to justify the price I would pay?

For the current stock's filings, quotes or valuation, the assistant needs separate up-to-date sources. This repository does not supply live prices, a proprietary Buffett valuation score, trading functionality or brokerage access.

For AI assistants and developers

Read AGENTS.md for the correct order of research, what files are available, how to cite original sources, and how to report missing evidence. The packaged skill instructions are in skills/warren-buffett-brain/SKILL.md.

You can search the historical index locally, without running an LLM or connecting to a service:

python3 scripts/search_knowledge.py "GEICO competitive advantage" --json

Python is for local scripts and the optional desktop MCP, not a requirement for reading the website or pasting the ChatGPT prompt.

From business quality to valuation (optional)

Understanding a company's moat does not establish that the stock is cheap. Continue the research in ValuEdge if you want to examine separate model-based valuations and assumptions. ValuEdge is a separate commercial research product operated by this project's maintainer; it is not needed to use Buffett Brain and its estimates are not Buffett's views.

Trust, rights, and contributions

This project is not affiliated with, endorsed by or operated by Warren Buffett or Berkshire Hathaway. Its original code and research summaries are MIT-licensed. Third-party books, shareholder letters, recordings, photographs, brands and other materials remain with their rights holders; this project links to them instead of redistributing full texts.

For source standards and corrections, see research and rights, contributing, privacy and terms. Report missing or misattributed evidence through GitHub Issues.

Maintainer-only deployment and submission details are in docs/MAINTAINER.md, not required for ordinary use.