Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add Mrjie7205/serenity-bottleneck-hunter --skill serenity-bottleneck-huntergit clone --depth 1 https://github.com/Mrjie7205/serenity-bottleneck-hunterWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/mrjie7205/serenity-bottleneck-hunter/serenity-bottleneck-hunter)<a href="https://agentmods.dev/skills/mrjie7205/serenity-bottleneck-hunter/serenity-bottleneck-hunter"><img src="https://agentmods.dev/badge/skills/mrjie7205/serenity-bottleneck-hunter/serenity-bottleneck-hunter/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/mrjie7205/serenity-bottleneck-hunter/serenity-bottleneck-hunter"><img src="https://agentmods.dev/badge/skills/mrjie7205/serenity-bottleneck-hunter/serenity-bottleneck-hunter.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00116 | $0.08554 |
| Opus 5 | $0.00058 | $0.04277 |
| Sonnet 5 | $0.00023 | $0.01711 |
| Haiku 4.5 | $0.00012 | $0.00855 |
Grade A, and why
serenity-bottleneck-hunter scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
curl "https://eodhd.com/api/search/<COMPANY_NAME>?api_token=$KEY&fmt=json" How it starts
The opening of the file, as written. The whole thing — 224 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Serenity Bottleneck Hunter
把一个投资主题转成一份被忽视的上游瓶颈候选股名单 + 论证 + 目标价/时间框架。核心不是抄 Serenity 的票,而是复用他的逻辑去选新股。
📓 每条纪律是怎么用一次真实翻车换来的,见
reference/lessons.md。本文件只讲"做什么";完整事故复盘在那边,用〔… → lessons.md#锚点〕指过去。
何时用
- 用户给出一个主题/趋势,想要"沿这个方向能买什么"。
- 用户给一个上游环节/材料,想找对应的瓶颈公司。
- ❌ 不要用于:已知标的的纯财务分析、抄作业式"他买了啥"。
核心理念
逆向拆解供应链,在机构与分析师发现之前,埋伏那个无人察觉的上游瓶颈,用催化剂兑现。 alpha 来自"早于机构发现主题",不是抄到最低点。
工作流(7 步)
Step 1 · 确认资本开支确定性 这个主题的钱为什么"一定"会花?规模、周期多长?需求确定性 > 个股故事性。先归类:Bottleneck(瓶颈)/ Disruption(颠覆)/ Evolution(演进)?(本 skill 主攻 Bottleneck)
- 标明需求来源:政府/国防(节奏慢、看订单与预算周期)还是商业(超大规模厂商 capex、节奏快)?二者估值锚与择时节奏不同,Step 6/7 据此调整。
Step 2 · 逆向拆链 + 广扫纪律(5 层产业链 + 跨主题 root + 可选第 6 层独有 IP) 列出 5 层 从下游到上游:下游对照 → 中游系统 → 中游器件 → 上游设备 → 上游材料/代工。跳过人人都盯的下游龙头。
-
必做"广扫供应商":除最上游材料层外,单独再搜一轮该主题的子系统/器件/卖铲子供应商,否则漏 ④ 原型标的。〔教训:商业航天漏 Redwire → lessons.md#redwire〕
-
广扫颗粒度:覆盖 5 层,每层至少 2-3 个标的。广扫不到位 ≠ 板块没机会 —— 停手前先问"是真没机会,还是没扫够"。〔教训:800VDC 6 只→15 只才挖出 STM → lessons.md#breadth-stm〕
-
跨主题 root 节点:ARM / CDNS / SNPS 放在 chain-viz 上方独立的"跨主题 root"小卡片,显示一次,不画进每个主题的产业链层(三家是所有芯片设计驱动主题共用的根)。〔教训:7 份报告 IP/EDA 层 100% 重复 → lessons.md#ip-eda-root〕
<div class="cross-theme-root"> <span class="label">跨主题 root(芯片设计驱动主题共用)</span> ARM ⭐⭐ + CDNS ⭐⭐ + SNPS ⭐⭐ </div>其 ⭐⭐ 状态来自
tracking/cross_theme_index_snapshot.csv,不每个主题重画。 -
可选第 6 层 · 主题独有 IP/EDA 玩家(仅当主题真有独立玩家才画):
- 自动驾驶 L4 → 真有独立 IP:MBLY(Mobileye 授权给 BMW/Audi/Geely)— 必须画
- 个人 AI PC → ARM 在 PC 是真单源(Apple M + Qualcomm Snapdragon X + AMD AI300)— 保留画在产业链顶层 + 标本主题真单源
- AI Agent / 物理 AI / AMR → 无独立 IP 玩家 → 不画第 6 层,root section 提一次即可
- MLCC / 800VDC / 商业航天 → 不依赖芯片设计,完全跳过 IP/EDA
- 凑数禁令:没有真独立玩家的主题不强制画第 6 层 —— IP/EDA 是 optional 不是 mandatory。
-
A 穷尽性 · 已知玩家全集:每个主题开扫前先列已知玩家全集(全球+全市场,涵盖上市/私有/被并购),逐一显式标记
covered/private(跳过+原因)/acquired(标合并实体)/delisted/untracked(EODHD/yfinance 无数据)。不允许"默认遗漏",不允许"我扫了 5 家就够了"。写在报告底部"已知玩家全集 audit"块。〔示例:L4 LiDAR 全集枚举 → lessons.md#player-census〕 -
A+ ETF audit(强制工具化兜底):A 项"凭记忆列全集"不够,会漏主仓玩家。 每个新主题开 forward_picks 扫描前,必跑
python scripts/theme_etf_coverage.py --etfs <ETF1>,<ETF2>,... --theme "<主题名>"→ 拉主题 ETF top 25 持仓、与 forward_picks diff、输出tracking/_etf_audit_<theme>.json。对每只 new candidate 人工判定产业链层 + 原型 + 是否纳入;top 25 持仓 100% 必 audit(主仓占权重 >70%,小权重 me-too 可跳)。〔教训:Dogfood #10 漏 PANW/CRWD 等 34 只、covered 率 50% → lessons.md#etf-audit〕- 主题 ETF 映射:
- AI Agent / SaaS / Software →
IGV+WCLD+AIQ - 网络安全 →
CIBR+BUG+HACK - 半导体 →
SOXX+SMH+PSI - AI 算力/GPU →
AIQ+BOTZ - 网络/算力基础设施 →
WCLD+CLOU+SKYY - 物理 AI 机器人 →
BOTZ+ROBO+IRBO - 自动驾驶 →
DRIV+IDRV - 电网/能源 →
GRID+XLU+URNM - 清洁能源/电气化 →
ICLN+LIT+KARS - 国防/航天 →
ITA+XAR+UFO - 5G / 通信 →
FIVG+NXTG - 数据中心 / 算力 →
DTCR+CLOU+SRVR
- AI Agent / SaaS / Software →
- 报告 audit section 必列:① ETF 清单 ② ETF top 持仓全集数 ③ covered/private/跳过原因逐一分类 ④ ETF coverage %。< 80% 必须诚实标"穷尽性纪律未达标"。
- 注:A 股无 stockanalysis ETF 持仓数据 → 改用申万行业指数成分 + 人工列已知玩家全集。
- 主题 ETF 映射:
-
A++ ticker 双向验证(强制):广扫前用 EODHD search 反查"代码 → 公司名"双向验证——A 项穷尽 + A+ ETF audit 都不防"错标"(标对公司名但 ticker 指向另一家)。每只候选必验:
curl "https://eodhd.com/api/search/<COMPANY_NAME>?api_token=$KEY&fmt=json" curl "https://eodhd.com/api/search/<TICKER>?api_token=$KEY&fmt=json"→ 两次返回的
Name必须一致。禁止凭记忆写 ticker,哪怕"我很确定"也必须 search。A 股尤其踩坑:6(沪市)/ 0(深市)/ 3(创业板)/ 688(科创板)/ 002(中小板)代码段交叉,同 5 位数字可能是不同公司。工具:scripts/ticker_truth.py(ground-truth 库)+scripts/verify_tickers.py(git pre-commit hook 自动拦截)。验证通过才写入 scan*.py。 〔教训:绿的谐波/信质电机/ASEKY/EHGO 共 4 个真错位 → lessons.md#ticker-verify〕 -
公司状态检查(写判定前必查):每只候选(含链路图非候选节点)搜一次当前状态——收购/被收购/私有化/IPO 上市/退市/破产重组(搜
<公司名> acquisition/IPO/merger)。收购中的标的 🔴 剔除(被收购反向验证选股方向、记进报告当佐证)。"私有/不可投"是状态性断言,LLM 的"某某私有"先验是训练快照、极易过期,必须当场搜证,状态写进 company_desc 带日期的 status 字段。〔教训:SkyWater 被 IonQ 收购 + SpaceX IPO 当天报告还标"私有" → lessons.md#company-status〕 -
私有公司诚实跳过:每个主题都有"真单源但私有/子部门"的标的。禁止为凑标的硬塞不纯/不可投的公司;在报告里显式列"已识别但跳过的瓶颈 + 原因"(本身就是高密度信号)。例:物理 AI 六维力传感器(坤维/字节灵犀/宇立全私)、鼎智科技(北交所 EODHD 不支持)→ 跳过。坦诚标"无干净纯 play"比硬凑更有价值。
-
主题边界声明:跑大类主题(物理 AI / AI 算力 / 能源转型)时,报告顶部必须界定本期边界 —— 大类下常有 3+ 个供应链差异显著的子领域,显式说"本期聚焦 X,其他子领域作为独立专题分批跑"。命名:
<大类>_<子领域>_完整分析报告.html。〔教训:物理 AI v1 默认=人形 → lessons.md#theme-scope〕
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 224 lines · 116 tokens per session scan A a3cc6a793f34
serenity-bottleneck-hunter is a skill published in the GitHub repository Mrjie7205/serenity-bottleneck-hunter (398 stars, last pushed 2mo ago), licensed MIT. It adds 116 tokens to every session and 8,554 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
sector-rotation
An analysis framework for comparing industries in the Chinese A-share stock market, using business conditions, price momentum, valuation, and money flows. It produces rankings and higher- or lower-allocation suggestions.
strategy-pivot-designer
Detect backtest iteration stagnation and generate structurally different strategy pivot proposals when parameter tuning reaches a local optimum.
twitter-reader
Read Twitter/X for financial research using opencli (read-only). Use this skill whenever the user wants to read their Twitter feed, search for financial tweets, view bookmarks, look up user profiles, or gather market sentiment from Twitter/X. Triggers include: "check my feed", "search Twitter for", "show my…
chenhao-limit-up
A framework for judging Chinese A-share stocks that have reached the daily price-rise limit, using market mood, sector leadership, and trading momentum.
furusato
A Japanese hometown-tax donation manager for furusato nozei, a system where donations to municipalities can qualify for an income-tax or local-tax deduction. It reads donation receipts, stores donation records, and calculates deduction limits.
reading-receipt
An image-reading workflow for extracting structured information from receipts, invoices, and hometown-tax donation certificates. It can first extract text from PDFs and otherwise read their images.