# How to Use the Industry-Research Skill for Supply Chain Analysis in AI-Berkshire

> Master supply chain analysis with AI-Berkshire's industry-research skill. Learn the 7-step pipeline to build investment logic, validate hypotheses, and generate portfolio recommendations.

- Repository: [Xbt Lin/ai-berkshire](https://github.com/xbtlin/ai-berkshire)
- Tags: how-to-guide
- Published: 2026-07-11

---

**The industry-research skill executes a seven-step structured pipeline that constructs investment logic chains, validates hypotheses, maps full industry panoramas, and generates portfolio recommendations for supply chain analysis.**

The **industry-research skill** in the `xbtlin/ai-berkshire` repository provides a comprehensive framework for supply chain analysis and industry value-chain scanning. Defined in [`skills/industry-research.md`](https://github.com/xbtlin/ai-berkshire/blob/main/skills/industry-research.md), this canonical workflow enables both Claude Code and Codex users to perform rigorous, data-driven assessments that map upstream, midstream, and downstream segments while identifying beneficiary companies across global markets.

## What Is the Industry-Research Skill?

The **industry-research skill** is the canonical workflow for full-stack industry and value-chain analysis within the AI-Berkshire ecosystem. Located at [`skills/industry-research.md`](https://github.com/xbtlin/ai-berkshire/blob/main/skills/industry-research.md), this file serves as the single source of truth for both Claude Code and Codex implementations, providing a structured methodology for deconstructing complex supply chains into analyzable components.

The skill is also packaged for Codex compatibility in [`codex-skills/industry-research/SKILL.md`](https://github.com/xbtlin/ai-berkshire/blob/main/codex-skills/industry-research/SKILL.md), allowing invocation across multiple interfaces while preserving the same underlying architecture and validation logic.

## The Seven-Step Supply Chain Analysis Pipeline

When you invoke the **industry-research skill** for supply chain analysis, the system executes a rigorous seven-step pipeline:

### 1. Investment-Logic Chain Construction

The workflow begins by drawing a **causal chain** from macro trends to beneficiary segments. This follows the Chinese investment framework: 底层趋势 (underlying trend) → 需求 (demand) → 瓶颈 (bottleneck) → 受益产业链 (beneficiary industry chain). This step establishes the theoretical foundation for why a specific supply chain segment will capture value.

### 2. Chain Validation

Each arrow in the causal chain is **questioned and verified** against evidence. The skill generates a validation table with columns for 环节 (segment), 核心假设 (core assumptions), 验证方式 (validation method), and 数据来源 (data sources), ensuring that every logical connection is empirically grounded before proceeding.

### 3. Full-Chain Panorama Mapping

The industry is decomposed into **upstream, midstream, downstream, and auxiliary segments** (上游-中游-下游-辅助环节). For each segment, the analysis captures:
- 商业模式 (business model)
- 毛利率区间 (gross margin range)
- 竞争格局 (competitive landscape)
- 壁垒类型 (barrier types)
- 周期性 (cyclicality)

### 4. Global Listed-Company Scan

Task agents systematically scan **all major markets** including 美股 (US stocks), A股 (A-shares), 港股 (Hong Kong stocks), and international listings. This produces a segment-wise table containing: 公司 (company), 代码 (ticker), 市值 (market cap), 一句话描述 (one-sentence description), 是否纯正标的 (purity indicator), and 所属环节 (segment classification).

### 5. Four-Master Analysis

For each Tier-1 and Tier-2 company identified, the skill runs the **"Four Masters" (四大师)** modules evaluating:
- 生意本质 (nature of business)
- 护城河 (moat/competitive advantage)
- 风险 (risks)
- 管理层 (management quality)
- 估值 (valuation)
- 推荐度 (recommendation rating)

### 6. Industry-Level Risk and Civilization Trend Assessment

The analysis incorporates **systemic risk checklists**, historical analogues, and the **Li Lu framework** (李录框架) for identifying long-term civilization trends that could impact the supply chain's structural viability.

### 7. Portfolio-Allocation Recommendation

The final step generates a **strategic allocation table** distinguishing between core positions (核心), satellite positions (卫星), options, and ETF configurations, complete with specific buy/sell signals based on the preceding analysis.

## Data Validation and Financial Rigor

All steps in the **industry-research skill** are **data-driven** and employ automated validation tools. According to [`skills/investment-research.md`](https://github.com/xbtlin/ai-berkshire/blob/main/skills/investment-research.md), the skill invokes [`tools/financial_rigor.py`](https://github.com/xbtlin/ai-berkshire/blob/main/tools/financial_rigor.py) for precise financial calculations and [`tools/report_audit.py`](https://github.com/xbtlin/ai-berkshire/blob/main/tools/report_audit.py) for multi-source cross-validation.

These utilities ensure that market-cap figures are verified, financial statements are cross-referenced against multiple sources, and data quality meets publication standards before the final report is generated.

## How to Invoke the Industry-Research Skill

You can trigger the **supply chain analysis workflow** using simple slash commands in Claude Code or through the Codex interface:

```bash

# Basic supply chain analysis

/industry-research 供应链

# Specific theme analysis (e.g., EV battery supply chain)

/industry-research "电动汽车电池供应链"

```

Upon execution, the skill automatically generates a comprehensive markdown report in the repository root following the **Output Requirements** section of [`industry-research.md`](https://github.com/xbtlin/ai-berkshire/blob/main/industry-research.md). The report filename follows the convention `~/[theme]-industry-[YYYYMMDD].md`, such as `~/电动汽车电池供应链-industry-20260711.md`.

## Summary

- The **industry-research skill** resides in [`skills/industry-research.md`](https://github.com/xbtlin/ai-berkshire/blob/main/skills/industry-research.md) and provides a canonical seven-step workflow for supply chain analysis.
- The pipeline progresses from **investment-logic construction** through **chain validation**, **panorama mapping**, **global company scanning**, **Four-Master analysis**, **risk assessment**, to **portfolio recommendations**.
- Validation tools [`tools/financial_rigor.py`](https://github.com/xbtlin/ai-berkshire/blob/main/tools/financial_rigor.py) and [`tools/report_audit.py`](https://github.com/xbtlin/ai-berkshire/blob/main/tools/report_audit.py) ensure data accuracy and cross-verification throughout the process.
- Invocation is available via `/industry-research` commands in Claude Code or through the Codex package at [`codex-skills/industry-research/SKILL.md`](https://github.com/xbtlin/ai-berkshire/blob/main/codex-skills/industry-research/SKILL.md).
- Output generates as a structured markdown report containing tables, chain diagrams, and master-analysis sections.

## Frequently Asked Questions

### What file contains the canonical definition of the industry-research skill?

The canonical skill definition lives in [`skills/industry-research.md`](https://github.com/xbtlin/ai-berkshire/blob/main/skills/industry-research.md) within the `xbtlin/ai-berkshire` repository. This file serves as the source of truth for both Claude Code and Codex implementations. A generated wrapper for Codex compatibility is available at [`codex-skills/industry-research/SKILL.md`](https://github.com/xbtlin/ai-berkshire/blob/main/codex-skills/industry-research/SKILL.md).

### How does the skill validate financial data during supply chain analysis?

The skill calls [`tools/financial_rigor.py`](https://github.com/xbtlin/ai-berkshire/blob/main/tools/financial_rigor.py) for precise financial calculations and [`tools/report_audit.py`](https://github.com/xbtlin/ai-berkshire/blob/main/tools/report_audit.py) for automated data quality audits. These utilities perform market-cap verification, multi-source cross-validation, and ensure figures meet publication standards before inclusion in the final report.

### Can I use the industry-research skill outside of Claude Code?

Yes. While the skill is accessible via the `/industry-research` slash command in Claude Code, it is also packaged as a Codex skill in [`codex-skills/industry-research/SKILL.md`](https://github.com/xbtlin/ai-berkshire/blob/main/codex-skills/industry-research/SKILL.md). This allows invocation from any supported interface including direct CLI usage or Codex prompts, maintaining the same architectural logic across platforms.

### Which markets does the global listed-company scan cover?

The task agents scan four major market categories: 美股 (US stocks), A股 (A-shares), 港股 (Hong Kong stocks), and international listings. This comprehensive coverage ensures identification of pure-play and related companies across the entire supply chain regardless of geographic listing location.