# AI Berkshire

> A collection of Claude Code and Codex skills that structure value-investing research using the methods of Buffett, Munger, Duan Yongping and Li Lu.

AI Berkshire is an open-source collection of investment-research skills for Claude Code and Codex, published by xbtlin on GitHub. It packages the methodologies of Warren Buffett, Charlie Munger, Duan Yongping and Li Lu into 20 slash-command skills covering company research, earnings analysis, industry screening, portfolio management and thinking tools. The latest listed release is v1.0.0.

## What It Is

AI Berkshire is a skill framework that runs inside an AI coding client rather than a standalone application. Users install the skills into Claude Code (as commands) or Codex (as skills and optional slash prompts), then invoke them with a company, industry or report as input. Outputs are structured research reports with scores, pass/gray-zone/fail conclusions and price-range suggestions, written as Markdown files. The README states that the project is for study and research and is not investment advice.

## How the Research Workflow Works

Team-type skills such as `/investment-team` and `/earnings-team` use a Team Lead that dispatches four parallel agents, one per master's perspective. Each agent searches, cross-checks data and scores independently, and the lead then synthesizes the results. Lighter skills such as `/quality-screen` and `/news-pulse` call tools directly for quick screening or price-move attribution. The README describes built-in safeguards: an A/B/C information-richness rating, Munger-style inversion checks, a veto checklist, a contrarian check and an explicit "gray zone" label when data is thin.

## Skill Categories

The 20 skills are grouped into deep research (investment-research, investment-team, management-deep-dive, private-company-research, deep-company-series), earnings analysis (earnings-review, earnings-team), industry screening (industry-research, industry-funnel, quality-screen, bottleneck-hunter, era-alpha, investment-checklist), portfolio management (income-investment, portfolio-review, thesis-tracker, thesis-drift, news-pulse) and thinking tools (dyp-ask, financial-data, wechat-article).

## Financial Rigor Tooling

A Python tool, `tools/financial_rigor.py`, handles market-cap verification, valuation checks, multi-source cross-validation, three-scenario valuation, Benford's-law detection and a precise calculator. It uses Python `decimal.Decimal` rather than floats. The project also documents a cost note: deep research skills consume many tokens, and the maintainers suggest using lighter skills first to screen before running full analyses.

## Features
- 20 investment-research skills for Claude Code and Codex
- Multi-agent parallel research with four master perspectives
- Earnings report review and team-based analysis
- Industry chain scanning and funnel screening
- Buffett-style pre-purchase checklist
- Private company research with confidence labels
- Portfolio review and thesis tracking
- Thesis drift detection between reports
- Stock price move attribution (news-pulse)
- Decimal-precise financial verification tool
- Multi-source data cross-validation
- Install scripts for Claude Code and Codex

## Integrations
Claude Code, Codex

## Platforms
WINDOWS, MACOS, LINUX, API, CLI

## Pricing
Open Source

## Version
v1.0.0

## Links
- Website: https://github.com/xbtlin/ai-berkshire
- Documentation: https://github.com/xbtlin/ai-berkshire#readme
- Repository: https://github.com/xbtlin/ai-berkshire
- EveryDev.ai: https://www.everydev.ai/tools/ai-berkshire