Agent skill

Supply Chain Bottleneck Hunter

Scans a long-running industry trend for supply chain chokepoints, aiming to find second- and third-layer suppliers that the market has not yet priced in.

MITAuto-check passedBusiness, Finance & HR

SKILL.md written in Chinese; this summary is our English description.

Install Supply Chain Bottleneck Hunter

skills CLI

$ npx skills add xbtlin/ai-berkshire --skill bottleneck-hunter -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI

$ gh skill install xbtlin/ai-berkshire bottleneck-hunter --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy

$ git clone --depth 1 https://github.com/xbtlin/ai-berkshire.git skills-src && mkdir -p .claude/skills && cp -r skills-src/codex-skills/bottleneck-hunter .claude/skills/bottleneck-hunter && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Paste into Claude Code

Install the "bottleneck-hunter" agent skill from https://github.com/xbtlin/ai-berkshire/tree/main/codex-skills/bottleneck-hunter into .claude/skills/bottleneck-hunter/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "bottleneck-hunter", then confirm the skill loads.

Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

$skill-installer

$skill-installer install https://github.com/xbtlin/ai-berkshire/tree/main/codex-skills/bottleneck-hunter

Type this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.

skills CLI

$ npx skills add xbtlin/ai-berkshire --skill bottleneck-hunter -a codex

Project install goes to .agents/skills/; add -g for ~/.codex/skills/.

GitHub CLI

$ gh skill install xbtlin/ai-berkshire bottleneck-hunter --agent codex

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/xbtlin/ai-berkshire.git skills-src && mkdir -p .agents/skills && cp -r skills-src/codex-skills/bottleneck-hunter .agents/skills/bottleneck-hunter && rm -rf skills-src

Use ~/.agents/skills/ instead of .agents/skills for a personal install.

Codex skills documentation · loads skills from .agents/skills/

Paste into Codex

Install the "bottleneck-hunter" agent skill from https://github.com/xbtlin/ai-berkshire/tree/main/codex-skills/bottleneck-hunter into .agents/skills/bottleneck-hunter/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "bottleneck-hunter", then confirm the skill loads.

Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

skills CLI

$ npx skills add xbtlin/ai-berkshire --skill bottleneck-hunter -a cursor

Project install goes to .agents/skills/; add -g for ~/.cursor/skills/.

GitHub CLI

$ gh skill install xbtlin/ai-berkshire bottleneck-hunter --agent cursor

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/xbtlin/ai-berkshire.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/codex-skills/bottleneck-hunter .cursor/skills/bottleneck-hunter && rm -rf skills-src

Use ~/.cursor/skills/ instead of .cursor/skills for a personal install.

Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/

Paste into Cursor

Install the "bottleneck-hunter" agent skill from https://github.com/xbtlin/ai-berkshire/tree/main/codex-skills/bottleneck-hunter into .cursor/skills/bottleneck-hunter/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "bottleneck-hunter", then confirm the skill loads.

Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

gemini skills install

$ gemini skills install https://github.com/xbtlin/ai-berkshire.git --path codex-skills/bottleneck-hunter

--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.

skills CLI

$ npx skills add xbtlin/ai-berkshire --skill bottleneck-hunter -a gemini-cli

Project install goes to .agents/skills/; add -g for ~/.gemini/skills/.

GitHub CLI

$ gh skill install xbtlin/ai-berkshire bottleneck-hunter --agent gemini-cli

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/xbtlin/ai-berkshire.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/codex-skills/bottleneck-hunter .gemini/skills/bottleneck-hunter && rm -rf skills-src

Use ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.

Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/

Paste into Gemini CLI

Install the "bottleneck-hunter" agent skill from https://github.com/xbtlin/ai-berkshire/tree/main/codex-skills/bottleneck-hunter into .gemini/skills/bottleneck-hunter/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "bottleneck-hunter", then confirm the skill loads.

Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

GitHub CLI

$ gh skill install xbtlin/ai-berkshire bottleneck-hunter

Installs for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).

skills CLI

$ npx skills add xbtlin/ai-berkshire --skill bottleneck-hunter -a github-copilot

Project install goes to .agents/skills/; add -g for ~/.copilot/skills/.

Manual copy

$ git clone --depth 1 https://github.com/xbtlin/ai-berkshire.git skills-src && mkdir -p .github/skills && cp -r skills-src/codex-skills/bottleneck-hunter .github/skills/bottleneck-hunter && rm -rf skills-src

Use ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.

GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/

Paste into GitHub Copilot

Install the "bottleneck-hunter" agent skill from https://github.com/xbtlin/ai-berkshire/tree/main/codex-skills/bottleneck-hunter into .github/skills/bottleneck-hunter/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "bottleneck-hunter", then confirm the skill loads.

GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

skills CLI

$ npx skills add xbtlin/ai-berkshire --skill bottleneck-hunter -a opencode

OpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.

GitHub CLI

$ gh skill install xbtlin/ai-berkshire bottleneck-hunter --agent opencode

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/xbtlin/ai-berkshire.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/codex-skills/bottleneck-hunter .opencode/skills/bottleneck-hunter && rm -rf skills-src

Use ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.

OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/

Paste into OpenCode

Install the "bottleneck-hunter" agent skill from https://github.com/xbtlin/ai-berkshire/tree/main/codex-skills/bottleneck-hunter into .opencode/skills/bottleneck-hunter/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "bottleneck-hunter", then confirm the skill loads.

OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

Facts

Skill name: bottleneck-hunter
GitHub stars: 17k
Token cost: ~2.6k tokens
SKILL.md length: 646 words
Files: 1
Skills in repo: 22
Repo updated: Oct 8, 2026
First seen: Oct 7, 2026
Licence: MIT

At a glance

Scans a long-running industry trend for supply chain chokepoints, aiming to find second- and third-layer suppliers that the market has not yet priced in.

Works in 5 steps: AI基础设施建设 — 数据中心、GPU集群、网络互联、电力 → 能源转型 — 核电重启、电网升级、储能 → 国防现代化 — 西方军费上升周期、供应链重构 → …
Looking for supply chain bottlenecks behind a long-term industry trend
SKILL.md covers Codex adapter note, 核心理念, 第一步：超级趋势确认 and 第二步：供应链物理拆解, plus 5 more sections
Calls python3

What it does

The skill is written in Chinese and is a Codex adapter of a research workflow from an AI-era value-investing framework. Instead of asking which stocks an AI recommends, it asks which link in the supply chain runs short first if a trend keeps expanding. It argues that first-layer bottlenecks such as GPUs, high-bandwidth memory and power are already priced in, and looks for second- and third-layer items such as optical modules, lasers, substrates, wafers, test services and special glass fiber.

Step one confirms a trend using criteria for persistence, physical build-out, scale and demand outpacing supply, with a starting list covering AI infrastructure, energy transition, defense modernization, semiconductor reindustrialization and the space economy. Step two breaks the chain into layers down to physical components and uses web searches for bottleneck and shortage terms. The adapter notes tell the agent to confirm today's date first and state the data cutoff in the report header, cross-check financial data, use exact arithmetic tools and label uncertainty and source gaps.

When your agent uses it

Looking for supply chain bottlenecks behind a long-term industry trend
Breaking an industry chain into layers to find overlooked suppliers
Building a research report with a stated data cutoff date

Example prompts

“Run a bottleneck scan on AI infrastructure and list the second-layer suppliers.”
“Which parts of the space economy supply chain could run short first? Do the layer breakdown.”
“Scan the energy transition trend only and show where capacity lags demand.”

Requirements

Web search access for supply chain research
Python for the shared financial rigor tools in the repository

Workflow steps

5 steps, taken from the first numbered list in SKILL.md.

AI基础设施建设 — 数据中心、GPU集群、网络互联、电力
能源转型 — 核电重启、电网升级、储能
国防现代化 — 西方军费上升周期、供应链重构
半导体再工业化 — 美欧日补贴建厂、设备/材料瓶颈
太空经济 — 卫星互联网、发射频次激增

What it can do on your machine

Read from SKILL.md and the folder at commit a221a20. It shows what the files ask for, not the result of running them.

Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
Shell commands in SKILL.md call:
- python3
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Supply Chain Bottleneck Hunter loads about 2.6k tokens when it runs. Until then it costs about 25 tokens; SKILL.md has 646 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it: ~25
When it runs · the whole SKILL.md, loaded when a task matches: ~2.6k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from xbtlin/ai-berkshire at commit a221a20, republished under its MIT licence (© xbtlin). 646 words, ~2,630 tokens.

Download SKILL.mdSave it as .claude/skills/bottleneck-hunter/SKILL.md (or your agent's skills folder).

name: bottleneck-hunter
description: AI Berkshire skill: 供应链瓶颈猎手：AI驱动的全球产业链瓶颈套利. Source: skills/bottleneck-hunter.md.

Codex adapter note

This skill is generated from skills/bottleneck-hunter.md so Claude Code and Codex users share one canonical workflow.

Treat $ARGUMENTS as the user's request in the current Codex thread.
When the source mentions Claude-only surfaces such as Task, Agent, WebSearch, Bash, Read, or Write, use the closest Codex capability available in this session: subagents when available, web search when needed, shell commands for local tools, and normal file edits for workspace files.
Use shared project tools from tools/ in this repository. Prefer running commands from the repository root with paths like python3 tools/financial_rigor.py ...; if the current thread starts outside the repo, locate the actual checkout path first instead of assuming a fixed home-directory path.
Before starting research, run the date command to confirm today's date; treat it as the baseline for "latest" data and state the data cutoff date in the report header. Never assume the current date from training data.
Preserve the research quality rules from AGENTS.md: cross-check financial data, use exact arithmetic tools for valuation/math, and clearly label uncertainty and source gaps.

供应链瓶颈猎手：AI驱动的全球产业链瓶颈套利

对 $ARGUMENTS 超级趋势执行供应链瓶颈扫描与套利机会挖掘。

核心理念

不问"AI推荐什么股票"，问"如果这个趋势继续扩张，哪一环会先不够用？"

传统投研盯龙头、盯已知赛道。这个系统反过来：从物理供应链的咽喉位置出发，找那些没人注意但一旦断货整个行业都得停下来等的公司。

超额收益来源：第一层瓶颈（GPU、HBM、电力）已被充分定价。真正的alpha在第二层、第三层——光模块、激光器、InP衬底、SOI晶圆、外延设备、晶圆级测试、IC载板、特殊玻纤等。

第一步：超级趋势确认

1.1 趋势筛选标准

不在小风口里找幻觉，只追符合以下全部条件的超级趋势：

标准	要求	验证方法
持续性	至少3-5年确定性增长	搜索行业预测、资本开支计划
物理性	需要实际硬件/材料/设备建设	区分"软件升级"和"物理扩建"
规模性	全球资本开支>500亿美元/年	搜索头部玩家capex指引
加速性	需求增速>供给扩产速度	对比需求增长率vs产能扩张计划

1.2 当前跟踪的超级趋势清单

每次运行时更新，初始清单：

AI基础设施建设 — 数据中心、GPU集群、网络互联、电力
能源转型 — 核电重启、电网升级、储能
国防现代化 — 西方军费上升周期、供应链重构
半导体再工业化 — 美欧日补贴建厂、设备/材料瓶颈
太空经济 — 卫星互联网、发射频次激增

如果用户指定了具体趋势（如"AI基础设施"），只聚焦该趋势。

1.3 趋势验证输出

趋势名称：
核心驱动力：（一句话）
已发生的验证事件（至少3个）：
  1. [日期] [事件] [来源]
  2.
  3.
资本开支规模：全球约 $XX 亿/年，增速 YY%
供需缺口判断：需求增速 > 供给扩产速度？是/否/不确定
趋势确认：✅ 可追踪 / ❌ 证据不足，暂不追踪

第二步：供应链物理拆解

2.1 分层拆解框架

不停留在概念层，拆到物理实体。

Layer 0（终端）：最终产品/服务
    │
Layer 1（核心组件）：已被市场充分关注的核心硬件
    │                 ⬆ 定价充分，alpha有限
    │─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─
    │                 ⬇ 关注度低，alpha集中区
    │
Layer 2（子组件/材料）：支撑核心组件的零部件和材料
    │
Layer 3（上游设备/原料）：制造子组件所需的设备和原材料
    │
Layer 4（基础设施）：电力、冷却、土地、人才、认证

2.2 以AI基础设施为例的拆解模板

Layer 0：AI模型训练/推理服务
Layer 1：GPU/加速器、HBM内存、服务器、数据中心
Layer 2（重点扫描区）：
  ├─ 网络互联：光模块、光纤、交换芯片、铜缆
  ├─ 光通信核心：激光器（EML/VCSEL/CW）、调制器、光探测器
  ├─ 半导体材料：InP衬底、GaAs衬底、SOI晶圆、SiC衬底
  ├─ 先进封装：CoWoS载板、HBM TSV、ABF基板膜
  ├─ PCB/载板：高频高速PCB、IC载板、特殊玻纤布
  ├─ 测试：晶圆级测试（Probe Card）、老化测试、ATE
  ├─ 散热/冷却：液冷系统、CDU、浸没式冷却液
  └─ 电力连接：母线槽、UPS、配电柜、变压器
Layer 3：
  ├─ 外延设备：MOCVD、MBE
  ├─ 光刻/刻蚀：特殊波长光刻、InP刻蚀
  ├─ 原材料：高纯金属（铟、镓、锗）、特气、靶材
  └─ 认证/标准：MSA标准、Telcordia认证
Layer 4：
  ├─ 电力：核电、天然气发电、输变电
  ├─ 冷却水/散热基础设施
  └─ 数据中心土地/许可

2.3 其他趋势的拆解

对每个确认的超级趋势执行类似拆解。使用WebSearch搜索：

"{趋势} supply chain bottleneck 2026"
"{趋势} shortage critical component"
"{趋势} capacity constraint"
"{趋势} sole source supplier"

第三步：瓶颈识别——寻找"咽喉"

3.1 瓶颈判定6条标准

对Layer 2-3的每个环节，逐条评估：

#	标准	问题	得分
1	供给集中度	全球供应商≤3家？	🔴 ≤2家 / 🟡 3-5家 / 🟢 >5家
2	扩产周期	新增产能需要多久？	🔴 >2年 / 🟡 1-2年 / 🟢 <1年
3	替代难度	能否用其他技术/材料替代？	🔴 不可替代 / 🟡 部分可替代 / 🟢 易替代
4	产能利用率	当前产能利用率？	🔴 >90% / 🟡 70-90% / 🟢 <70%
5	需求增速	下游需求增速？	🔴 >50%/年 / 🟡 20-50% / 🟢 <20%
6	客户验证周期	新供应商进入需要多久验证？	🔴 >1年 / 🟡 6-12月 / 🟢 <6月

瓶颈评级：

🔴🔴🔴 ≥4个 → S级瓶颈（单点故障级，最高优先级）
🔴🔴 3个 → A级瓶颈（严重受限）
🔴 1-2个 → B级瓶颈（有压力但可控）
无🔴 → 非瓶颈，跳过

3.2 瓶颈地图输出

供应链瓶颈地图 — {趋势名称}
更新日期：YYYY-MM-DD

S级瓶颈（单点故障）：
  1. [环节名] — [原因一句话] — 供应商：[公司列表]
  2.

A级瓶颈（严重受限）：
  1.
  2.

B级瓶颈（有压力）：
  1.
  2.

最近变化（vs上次扫描）：
  - [新增/升级/降级/解除] [环节名] — [原因]

第四步：公司筛选——从瓶颈到标的

4.1 对每个S级和A级瓶颈，找出所有相关上市公司

搜索方式：

WebSearch "{瓶颈环节} supplier listed company"
WebSearch "{瓶颈环节} manufacturer stock"
WebSearch "{瓶颈产品} market share company"

4.2 初筛标准（快速过滤）

标准	要求	理由
上市状态	已上市（A/港/美/日/台/欧）	可交易
瓶颈业务占比	>30%收入来自瓶颈环节	纯正度
市值	优先<100亿美元	大市值已被充分定价
流动性	日均成交额>100万美元	可进出

4.2.1 估值检查（必须执行，不可跳过）

瓶颈真实 ≠ 投资机会。 必须对每家公司计算 PS、PE 并标注在报告中。以下用组合条件判断估值是否透支：

估值红灯（满足任意一条 → 信号强度封顶 ★★，标注"⚠️ 估值透支"）

市值 > TAM的20%：公司市值已超过其可触达市场规模的20%，说明增长预期已过度内化
PS > 30x 且收入增速 < 100%：高估值但增速不足以支撑。增速>100%的公司可豁免PS红线，但仍需标注"⚠️ 高估值需高增速持续验证"
市值 > 5年乐观收入预测的10倍：即使最乐观假设全部兑现，当前定价仍然过高
增发后60天内股价翻倍：情绪驱动特征明显，信号强度降一级

估值黄灯（需要额外解释，否则降级）

亏损 + PS > 15x：允许进入 ★★★ 但必须说明盈利路径和时间表
PS 是同行盈利公司的 5 倍以上：必须解释溢价来源（市场份额、增速差、壁垒差异）
PE > 80x：需要计算 PEG 并说明增速是否支撑

估值绿灯（加分项）

PS < 10x 且收入在增长 → 信号强度可加一级
PE < 30x 且有护城河 → 标注"估值具有安全边际"

Show full SKILL.md (257 more words)Show less

估值合理性检验（必做）

对每家标的回答："以当前市值买入，假设最乐观情景全部兑现，10年后以25x PE退出，年化回报是多少？" 年化回报<10%→标注"当前价格不具备安全边际"。

注意：估值检查的目的是防止推荐"PS 100倍的亏损公司"这种明显错误，而不是排除所有高估值早期公司。关键是增速、TAM、竞争格局能否支撑当前估值，需要具体分析而非一刀切。

台股标的取数：供应链扫描常命中台股（4位数字代码，如 2330、3661），其行情/PER/市值/月营收统一用 python3 tools/twstock_data.py（FinMind 数据源，用法见 skills/financial-data.md 台股章节）。台股每月10日前强制披露上月营收，月营收同比是验证瓶颈"量价齐升"最快的公开信号。

4.3 深度筛选维度

对通过初筛的公司，逐家评估：

## {公司名}（{代码}）

**瓶颈定位**：
- 在供应链中的具体位置
- 市场份额：全球第X，占比XX%
- 客户名单（已知的）

**产能与扩张**：
- 当前产能 / 利用率
- 扩产计划 / 时间表
- 扩产所需资金 vs 现有现金

**财务快照**：
- 市值 / 收入 / 利润 / 增速
- 瓶颈业务收入占比
- 毛利率趋势（瓶颈越紧，毛利率应上升）

**风险检查清单**：
- [ ] 替代技术风险：能否被绕过？
- [ ] 稀释风险：有无大量增发/可转债？
- [ ] 地缘风险：是否在敏感地区/受出口管制？
- [ ] 管理层风险：有无不良记录？
- [ ] 客户集中风险：是否过度依赖单一客户？
- [ ] 估值透支：当前估值是否已反映3年增长？

**瓶颈持续性判断**：
- 这个瓶颈什么时候会被解除？
- 解除后这家公司还有什么？
- 是一次性还是持续的？

第五步：交叉验证——不只听一个故事

5.1 正向验证

验证项	问题	搜索方式
客户验证	头部客户是否已签约/导入？	搜索公司公告、客户财报提及
收入验证	瓶颈是否已体现在收入增长中？	搜索最近2-3个季度财报
价格验证	产品是否在涨价？	搜索行业报价、分析师报告
产能验证	产能是否真的紧张？	搜索交期数据、客户抱怨
资本验证	是否有扩产资本开支？	搜索公司capex指引

5.2 反向验证（芒格式否定）

反向问题	意义
聪明人为什么不买这只股票？	寻找已知的bearish论点
这个瓶颈能被绕过吗？有什么替代路线？	技术路线风险
中国/其他玩家能不能很快复制产能？	供给冲击风险
终端需求如果放缓50%，这家公司会怎样？	下行敏感度
管理层过去有没有在高点增发稀释？	管理层信任度
当前估值隐含了什么增长假设？	估值合理性

5.3 信号交叉验证

同一瓶颈的多家公司是否都在涨？（行业验证）
下游客户是否在财报中提到供应紧张？（客户验证）
行业协会/研究机构是否有相关数据？（第三方验证）

第六步：输出——瓶颈机会看板

6.1 瓶颈机会排名表

排名	公司	代码	市值	年收入	PS	PE	瓶颈环节	瓶颈评级	市场份额	收入增速	信号强度	估值判断
1					x	x		S/A			★1-5	合理/偏高/透支

必填字段：市值、年收入、PS、PE 为必填项，不可用"待核实"跳过。如果无法获取财务数据，信号强度不得超过 ★★。

信号强度评级（估值检查结果直接影响评级）：

★★★★★ 多重交叉验证、客户已导入、收入已体现、估值绿灯（PS合理+盈利或接近盈利）
★★★★ 大部分验证通过、估值绿灯或黄灯（需附解释）
★★★ 逻辑成立但部分待验证、估值黄灯可接受（如高增速早期公司）
★★ 早期信号，或瓶颈逻辑成立但估值红灯（市值>TAM 20%、PS>30x且增速不足、市值远超5年预测等）
★ 纯概念、未验证

6.2 每个机会的一页纸摘要

🎯 {公司名}（{代码}）— {一句话瓶颈定位}

为什么是瓶颈：
（2-3句话解释这个环节为什么卡脖子）

为什么是这家公司：
（2-3句话解释为什么是这家而不是别家）

催化剂时间线：
- 近期（1-3月）：[具体事件，如财报、产能投产、客户验证]
- 中期（3-12月）：[行业趋势、扩产节点]

主要风险：
1.
2.

关键数据：市值 $XX / 年收入 $XX / PS Xx / PE Xx / 收入增速 XX% / 瓶颈业务占比 XX%

估值安全边际检验：以当前市值买入、10年后25x PE退出，需要净利润达到$XX，对应年收入$XX（是今天的X倍），年化回报XX%。结论：有/无安全边际。

交叉验证状态：✅ 客户验证 / ✅ 收入验证 / ✅ 估值合理 / ⚠️ 估值透支 / ❌ 未验证项

结论：值得深入研究 / 加入观察名单 / 暂不追踪

6.3 行动建议

标的	建议动作	理由
A	执行 `/investment-team` 深入研究	S级瓶颈+多重验证
B	加入观察名单，等下季财报	逻辑成立但收入未体现
C	暂不追踪	替代技术风险过高

第七步：存量更新——瓶颈地图动态维护

7.1 每次运行时的增量更新

检查已识别瓶颈是否仍然成立
- 有无新供应商进入？
- 产能是否已扩张到解除瓶颈？
- 替代技术是否有突破？
扫描新出现的瓶颈
- 搜索最近7天的supply chain / shortage / bottleneck新闻
- 检查财报季中的供应链相关disclosure
更新瓶颈评级（升级/降级/解除）

7.2 状态文件

在 reports/bottleneck-map/ 目录维护：

master-map.md — 瓶颈总地图（持续更新）
watchlist.md — 观察名单（持续更新）
YYYY-MM-DD/ — 每天一个文件夹，包含该天所有扫描报告
deep-dive/ — 深入分析的公司单独建文件

每小时扫描模式（定时任务使用）

每小时执行一次，采用"只在有料时出报告"的模式：

扫描流程（每小时）

新闻扫描：搜索过去1-2小时的供应链相关新闻
- 关键词：supply chain bottleneck, shortage, capacity constraint, allocation, lead time, sole source, 瓶颈, 缺货, 产能, 涨价
- 覆盖：英文+中文源
市场信号：检查已跟踪公司的股价变化（特别关注异常波动>5%）
财报/公告：检查是否有瓶颈相关公司发布财报或重大公告
估值机会：检查watchlist中公司是否有因大盘下跌等原因进入买入区间
判断是否出报告：
- 有新瓶颈信号、有明确标的机会、有重大状态变化 → 出报告
- 无新发现 → 不出报告，仅在日志中记录"本轮无新信号"

报告输出规则

每天一个文件夹：reports/bottleneck-map/YYYY-MM-DD/

文件命名规则（通过文件名一眼看出有没有标的）：

情况	文件名格式	示例
发现明确标的	`HH-MM-标的代码1-标的代码2.md`	`09-00-FORM-IBDN.md`
有瓶颈信号但无明确标的	`HH-MM-信号扫描.md`	`14-00-信号扫描.md`
无新发现	不生成文件	—

文件名中的标的代码 = 通过估值检查、值得深入研究的公司。 仅信号扫描阶段出现但估值不合格的公司不放在文件名中。

报告模板（有标的时）

markdown

# 瓶颈猎手 — YYYY-MM-DD HH:MM

## 明确标的

### {公司名}（{代码}）— {一句话瓶颈定位}

**为什么现在值得关注**：（触发本次关注的具体事件/数据变化）

**瓶颈定位**：Layer X，{环节名}，瓶颈评级 S/A/B
**财务快照**：市值 $XX / 年收入 $XX / PS Xx / PE Xx / 收入增速 XX%
**估值检查**：红灯/黄灯/绿灯（具体说明）
**估值安全边际**：10年25xPE退出法，年化回报XX%

**看多逻辑**（2-3条）：
1.
2.

**看空逻辑**（2-3条）：
1.
2.

**建议**：执行深入研究 / 加入观察 / 等待更好价格

---

## 其他信号（无明确标的）

| 环节 | 信号 | 来源 | 初步判断 |
|------|------|------|---------|

## 观察名单状态变化

（升级/降级/新增/移除，无变化则写"无变化"）

报告模板（仅信号扫描时）

markdown

# 瓶颈猎手信号扫描 — YYYY-MM-DD HH:MM

## 新信号

| 环节 | 信号描述 | 来源 | 是否有可投资标的 | 下一步 |
|------|---------|------|----------------|-------|

## 观察名单状态

无变化 / 有变化（列出）

AI研究偏见自觉

偏见	表现	应对
龙头偏好	搜索结果被大市值公司占据	刻意搜索小市值供应商、加"small cap"关键词
英文偏好	遗漏日韩台公司	必须搜索日/韩/台湾市场的供应商
叙事偏好	被"AI概念"标签吸引	只看实际供应链位置，不看市场标签
确认偏见	找到瓶颈后只找正面证据	强制执行反向验证（第五步）
时效偏见	依赖过时信息	优先搜索最近30天的数据

核心原则（最高优先级）

不让AI推荐股票，让AI拆供应链 — 问题比答案重要
物理优先 — 只关注需要实际物理产品/材料/设备的环节
第二层、第三层 — 不追已被充分定价的龙头
交叉验证 — 每个结论至少2个独立信源
诚实面对不确定性 — 找不到数据就写"数据不足"，不用推测填充
瓶颈有时效性 — 每个瓶颈都会被解除，关键是判断时间窗口
小市值≠好机会 — 小市值也可能是烂公司，必须过财务质量关
瓶颈真实≠投资机会 — 一家公司可以坐在最紧的瓶颈上，但如果PS>30x或仍在亏损，当前价格就不是买点。估值是硬门槛，不可被瓶颈纯正度、信号强度或叙事吸引力覆盖。 宁可错过一个涨了的瓶颈股，也不要在100倍PS买入一家亏损公司
遵循CLAUDE.md客观性原则 — 不预设看多，先数据后结论

输出要求

报告位置：
- 完整扫描：reports/bottleneck-map/{趋势名}-bottleneck-{YYYYMMDD}.md
- 每日扫描：reports/bottleneck-map/daily/{YYYY-MM-DD}-{am/pm}.md
- 瓶颈总地图：reports/bottleneck-map/master-map.md
- 观察名单：reports/bottleneck-map/watchlist.md
语言：中文
风格：直接、犀利、不说废话
数据：所有数据标注来源；估计值标"估计"
不预设立场：先摆数据 → 推逻辑 → 出结论
正反两面：每个核心判断附反面论据

Files

Just SKILL.md in codex-skills/bottleneck-hunter of xbtlin/ai-berkshire.

Open the folder on GitHubat commit a221a20

Compare with similar skills

Supply Chain Bottleneck Hunter next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Supply Chain Bottleneck Hunter compared with similar skills
Skill	Stars	Used in	Tokens	Auto-check	Licence	Repo updated
Supply Chain Bottleneck Hunter this skillxbtlin/ai-berkshire	17k	—	~2.6k	Automated safety check: Pass	MIT	yesterday
Eastmoney Market DataHKUDS/Vibe-Trading	35k	—	~1k	Automated safety check: Pass	MIT	yesterday
Stock Deep Analysis Workflowwbh604/UZI-Skill	7.1k	—	~9.1k	Automated safety check: Notes	MIT	1 mo ago
SEC EDGAR Filings FetcherHKUDS/Vibe-Trading	35k	—	~1.4k	Automated safety check: Pass	MIT	yesterday
A-Share Daily Reviewqusong0627/QuantMind	1.7k	—	~1.9k	Automated safety check: Pass	AGPL-3.0	today
Futu OpenAPI Market and Trading Assistantqusong0627/QuantMind	1.7k	—	~3.3k	Automated safety check: Notes	AGPL-3.0	today

Similar skills

Eastmoney Market Data
HKUDS/Vibe-Trading
Index of Eastmoney's free, no-token market data interfaces for China A-shares and Hong Kong stocks: fund flows, dragon-tiger lists, margin trading, reports and news.
35k GitHub stars~1k tokensUpdated yesterday
Business, Finance & HRAuto-check passed
Stock Deep Analysis Workflow
wbh604/UZI-Skill
Runs a staged deep analysis of a single stock on China A-share, Hong Kong and US markets, ending in an HTML report with valuation models and investor-panel scores.
7.1k GitHub stars~9.1k tokensUpdated 1 mo ago
Business, Finance & HRAuto-check: notes
SEC EDGAR Filings Fetcher
HKUDS/Vibe-Trading
Fetches U.S. SEC EDGAR data: resolves tickers to CIK numbers, lists recent 10-K, 10-Q and 8-K filings with document URLs, and pulls XBRL financial series.
35k GitHub stars~1.4k tokensUpdated yesterday
Business, Finance & HRAuto-check passed
A-Share Daily Review
qusong0627/QuantMind
Produces a post-market review report for the China A-share market from local QuantDB data, news sentiment and model signals, ending in a next-day direction call.
1.7k GitHub stars~1.9k tokensUpdated today
Business, Finance & HRAuto-check passed
Futu OpenAPI Market and Trading Assistant
qusong0627/QuantMind
Queries Futu quotes, options, fundamentals and accounts and places orders through the Futu OpenAPI Python SDK, defaulting to simulated trading.
1.7k GitHub stars~3.3k tokensUpdated today
Business, Finance & HRAuto-check: notes
US Market Data Toolkit
Geeksfino/finskills
Free Python scripts that fetch US stock data, SEC filings, insider trades and macro indicators, and run financial score calculators and portfolio analytics.
283 GitHub stars~1.2k tokensUpdated 7 mo ago
Business, Finance & HRAuto-check passed

Questions about Supply Chain Bottleneck Hunter

What does Supply Chain Bottleneck Hunter do?

Scans a long-running industry trend for supply chain chokepoints, aiming to find second- and third-layer suppliers that the market has not yet priced in. The skill is written in Chinese and is a Codex adapter of a research workflow from an AI-era value-investing framework. Instead of asking which stocks an AI recommends, it asks which link in the supply chain runs short first if a trend keeps expanding.

When should I use Supply Chain Bottleneck Hunter?

Supply Chain Bottleneck Hunter fits situations like: looking for supply chain bottlenecks behind a long-term industry trend; breaking an industry chain into layers to find overlooked suppliers; building a research report with a stated data cutoff date.

How do I install Supply Chain Bottleneck Hunter in Claude Code?

Run `npx skills add xbtlin/ai-berkshire --skill bottleneck-hunter -a claude-code`. Or copy the skill folder (codex-skills/bottleneck-hunter in xbtlin/ai-berkshire) into .claude/skills/bottleneck-hunter in your project. Claude Code loads it when a task matches its description.

How do I install Supply Chain Bottleneck Hunter in Codex?

Run `npx skills add xbtlin/ai-berkshire --skill bottleneck-hunter -a codex`. Or copy the skill folder (codex-skills/bottleneck-hunter in xbtlin/ai-berkshire) into .agents/skills/bottleneck-hunter in your project. Codex loads it when a task matches its description.

Can I use Supply Chain Bottleneck Hunter in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add xbtlin/ai-berkshire --skill bottleneck-hunter -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/bottleneck-hunter, .gemini/skills/bottleneck-hunter, .github/skills/bottleneck-hunter and .opencode/skills/bottleneck-hunter in your project.

What does Supply Chain Bottleneck Hunter need to run?

Going by SKILL.md and its folder, Supply Chain Bottleneck Hunter needs the command-line tools its instructions call (python3). Our summary lists: Web search access for supply chain research; Python for the shared financial rigor tools in the repository.

Does Supply Chain Bottleneck Hunter access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Supply Chain Bottleneck Hunter safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Supply Chain Bottleneck Hunter use?

Supply Chain Bottleneck Hunter is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Supply Chain Bottleneck Hunter use?

About 2.6k tokens (SKILL.md is roughly 11k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Supply Chain Bottleneck Hunter?

Skills that share tags, products or a category with Supply Chain Bottleneck Hunter: Eastmoney Market Data (HKUDS/Vibe-Trading, 35k stars), Stock Deep Analysis Workflow (wbh604/UZI-Skill, 7.1k stars), SEC EDGAR Filings Fetcher (HKUDS/Vibe-Trading, 35k stars) and A-Share Daily Review (qusong0627/QuantMind, 1.7k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Supply Chain Bottleneck Hunter?

xbtlin (a GitHub user) maintains it in xbtlin/ai-berkshire, which has 16,664 GitHub stars. The repository holds 22 skills in this directory. The repository was last updated on October 8, 2026.

Source: xbtlin/ai-berkshire on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.

Install Supply Chain Bottleneck Hunter

Facts

At a glance

What it does

When your agent uses it

Example prompts

Requirements

Workflow steps

What it can do on your machine

Tool permissions

Runs code

Network

Credentials

Context cost

Safety

SKILL.md

Codex adapter note

供应链瓶颈猎手：AI驱动的全球产业链瓶颈套利

核心理念

第一步：超级趋势确认

1.1 趋势筛选标准

1.2 当前跟踪的超级趋势清单

1.3 趋势验证输出

第二步：供应链物理拆解

2.1 分层拆解框架

2.2 以AI基础设施为例的拆解模板

2.3 其他趋势的拆解

第三步：瓶颈识别——寻找"咽喉"

3.1 瓶颈判定6条标准

3.2 瓶颈地图输出

第四步：公司筛选——从瓶颈到标的

4.1 对每个S级和A级瓶颈，找出所有相关上市公司

4.2 初筛标准（快速过滤）

4.2.1 估值检查（必须执行，不可跳过）

估值红灯（满足任意一条 → 信号强度封顶 ★★，标注"⚠️ 估值透支"）

估值黄灯（需要额外解释，否则降级）

估值绿灯（加分项）

估值合理性检验（必做）

4.3 深度筛选维度

第五步：交叉验证——不只听一个故事

5.1 正向验证

5.2 反向验证（芒格式否定）

5.3 信号交叉验证

第六步：输出——瓶颈机会看板

6.1 瓶颈机会排名表

6.2 每个机会的一页纸摘要

6.3 行动建议

第七步：存量更新——瓶颈地图动态维护

7.1 每次运行时的增量更新

7.2 状态文件

每小时扫描模式（定时任务使用）

扫描流程（每小时）

报告输出规则

报告模板（有标的时）

报告模板（仅信号扫描时）

AI研究偏见自觉

核心原则（最高优先级）

输出要求

Files

Compare with similar skills

Similar skills

More from xbtlin/ai-berkshire

Related

Questions about Supply Chain Bottleneck Hunter