zadanthony/serenity-skill

This repository contains an AI agent skill designed to apply a specific investment analysis methodology to US stocks, particularly in the AI supply chain, semiconductors, optical components, and adjacent sectors. The methodology being encoded belongs to a public investor on X (formerly Twitter) who goes by the name Serenity. The skill is a third-party project: the author built it by systematically organizing 2,071 of Serenity's public posts into a structured framework, and the project has no affiliation with or endorsement from Serenity. The disclaimer at the top of the README is clear that nothing produced by this skill is investment advice. The core idea behind the methodology is to trace capital flows along supply chains rather than asking whether a stock is worth buying outright. The approach looks for the smallest company in a supply chain that controls a critical bottleneck, where the market has not yet recognized how essential that position is. The phrase used is: not buying shovels, but finding the person who controls the supply of shovels. When you invoke the skill on a specific stock or sector, it is designed to produce a structured report. That report starts by explaining the relevant technology or business in plain language, then goes through each company and flags which positive criteria they meet and which red flags apply. It produces bear, base, and bull valuation ranges with explicit assumptions tied to each. Data cited in the report is labeled by its source type: whether it comes from a primary document, a management claim, an inference, or speculation. Risk factors and what would disprove the thesis are required sections. The skill consists of two files: SKILL.md, which is the entry point that defines the analysis pipeline and criteria, and methodology.md, which contains the full knowledge base derived from Serenity's public posts. To use it, you clone the repository into the skills directory of a compatible AI agent runtime and invoke it with a command like /serenity followed by a stock ticker or sector question. It works best when the host system has live internet access for pulling current financial data. The README is written primarily in Chinese with an English summary paragraph.