serenity-stock-choke: a prompt-based A-share supply chain chokepoint screener

The problem: A-share sector research has no standard order of operations

Most A-share sector notes start with the ticker. Someone hears that optical modules are moving, opens a watchlist, and works backwards to a story that justifies the price. serenity-stock-choke inverts that order. It is a reasoning scaffold: given a sector name, it walks you from demand shock to supply chain layer to the specific listed company sitting on the constrained node. The README frames the target as the node where a shortage would shake a trillion-yuan industry, and argues that the small-cap company controlling that node has the most elasticity. The audience is the discretionary A-share investor who already reads supply chain news and wants a consistent checklist rather than a tip. It is not a quant tool, and the README never presents it as one.

The Hormuz analogy and the three-step causal chain

The README's central analogy is the Strait of Hormuz: whoever owns a share of the strait owns pricing power when it closes. Translated to A-shares, the question becomes which segment is the Hormuz of a given sector, and which listed company controls it. The causal chain has three links. First, an AI boom, a policy push, or a technology jump creates demand. Second, upstream hardware and raw material demand surges and some supply chain links cannot keep pace. Third, the bottleneck link gains pricing power, and small caps on that link move hardest. The README illustrates the thesis with US-listed names from the original Serenity track record, including AXT Inc (AXTI.US) in InP substrates and Applied Optoelectronics (AAOI.US) in CPO lasers. Those are the author's cited examples, not results this repository produced, and the README does not publish a method for reproducing them.

The six-step analysis chain and what each step outputs

SKILL.md is described as the main prompt containing a six-step reasoning chain plus a toolbox. The steps are sequential. Step one locates the sector cycle and asks whether the driver is demand explosion, technology migration, or supply restriction. Step two traces the supply chain and asks which layer holds the bottleneck, producing a supply chain layer map. Step three finds A-share candidates sitting on that node and produces what the README calls a four-dimension signal card. Step four is the filter: separating genuine bottlenecks from companies riding the theme. Step five compares bullish and bearish signals in a table. Step six emits a structured report with operational suggestions and risk warnings. The design choice worth noting is that steps one through five are analytical and step six is prescriptive. The README also carries a risk note that logic verification can take one to three years, which sits awkwardly beside a report format that includes operational suggestions.

Installing serenity-stock-choke and running a first analysis

The repository has no package, no release, and no dependency manifest, so there is nothing to pip install. The README describes a skill that runs inside WorkBuddy, and the trigger is natural language rather than a command. Clone the repository to get SKILL.md and the references directory onto your machine, then load the skill in your host as you would any other skill.

git clone https://github.com/fadewalk/serenity-stock-choke
cd serenity-stock-choke
ls

The listing should show LICENSE, README.md, SKILL.md, references/ and scripts/. Once the skill is available, the README gives these trigger phrasings. Note that the examples are in Chinese, because the project targets A-share sectors.

用 serenity-stock-choke 分析 [板块名]
例：用 serenity-stock-choke 分析军工板块

The README lists two data tools the skill expects to call: neodata-financial-search for sector quotes, research reports and capital flows, and westock-data for chip distribution, block trades, margin financing and institutional ratings. Neither is bundled here; the README assumes they exist in your environment. If they are absent, the skill can still produce the six-step narrative, but every quantitative claim in it would be unsourced.

Where the framework breaks: no data, no verification, no backtest

The most concrete limitation is stated by the README itself: supply chain information is hard to verify and needs independent confirmation. The skill does not solve that. It supplies questions, not evidence. If you point it at a sector where the bottleneck is genuinely contested, the six steps will still produce a confident-looking report, because the output format does not distinguish between a node you can name a supplier for and one you inferred. There is also no backtest, no historical hit rate, and no release history in the repository. The US track record table is attributed to the original Serenity, not to this adaptation, and the README does not claim the A-share version reproduces it. The small-cap emphasis compounds the problem: the README warns that single-day moves of 20 to 30 percent are not rare in that segment, which means the elasticity argument and the drawdown argument are the same argument. Finally, the workflow is a poor fit for anyone who wants a ranked list. It returns a structured narrative about one sector at a time, and the sector tables in the README are marked as reference examples, not as output the tool generates.

serenity-stock-choke versus a conventional A-share screener

The natural alternative is a quantitative screener such as a factor library or a market-data terminal with a filter builder. The difference in approach is stark. A screener starts from a universe and applies numeric filters, so its output is reproducible and its failures are measurable. serenity-stock-choke starts from a narrative about demand and works down to a company, so its output is an argument you have to audit yourself. A screener can tell you which companies in a sector have a given market cap band and revenue concentration; it cannot tell you that a specific deposition step is the constraint. Conversely, the prompt framework can raise that hypothesis before you have written a single filter, but it cannot rank ten candidates by it. The two are complementary rather than competing, and the README's own reliance on neodata-financial-search and westock-data suggests the author expects you to bring a data terminal to the conversation.

Maintenance, licence, and what a fork commits you to

The repository is MIT licensed, which permits commercial use, modification and redistribution provided the copyright notice and permission notice are retained. Nothing here is legal advice; read LICENSE for the operative text. Maintenance is the weaker signal. The last push was on 2026-06-28, which is under three months before today, and the repository is not archived, but there are no retrieved releases and no version tags, so upgrades are not something you schedule. You pull main and diff. The practical cost is that SKILL.md and references/user_guide.md are prompt text, and prompt text drifts with the host model and the host skill format. A change in either can alter output without any commit in this repository. Pinning a commit hash is the only way to know what you are running, and the README does not document rollback or versioning.

Editorial conclusion

Adopt serenity-stock-choke if you already follow A-share supply chains and want a repeatable written frame for locating bottleneck nodes before you open a terminal. Do not adopt it if you expect a screener, a backtest, or a data feed: the repository contains a SKILL.md prompt, a README, a references/user_guide.md and a scripts directory, and the README states the skill is for reference only and is not investment advice. Verify first that WorkBuddy is the host you actually use, since the README's trigger examples are written for it, and read references/user_guide.md before trusting any sector table.

Frequently asked questions

Official sources

1. fadewalk/serenity-stock-choke on GitHub
2. Issues
3. License: MIT
4. README