muxuuu/serenity-skill

Serenity.skill is an AI research assistant for investors who see hot market themes, like AI chips, robotics, or new energy, flash across their screens but struggle to figure out which specific stocks or funds are actually worth investigating. It works as a "skill" you install into AI agent tools like Claude Code or OpenAI-compatible clients, giving the AI a structured methodology to break down hype into ranked research priorities. The project is inspired by the public research approach of a finance commentator known as Serenity on social media. The core idea is that in any booming sector, the real opportunity often hides in "bottleneck" points, parts of the supply chain that are hardest to scale or replace. When you ask the AI to research a topic, the skill guides it through a specific sequence: first, break the hot theme into its actual industry chain (from end-user demand down to materials and equipment). Then, identify which links in that chain have low supplier counts, long verification cycles, or high barriers to expansion. Finally, cross-reference public filings, earnings reports, and customer certifications to rank which companies sit closest to those bottlenecks versus which ones are just riding the hype. The skill requires the AI to ground its conclusions in real evidence like exchange filings and patents rather than social media chatter. This tool is built for retail investors and fund researchers facing information overload. If you see everyone talking about robotics but can't tell whether the real value is in finished machines, sensors, or gear reducers, this skill helps the AI systematically compare those segments. It also helps pressure-test specific claims, if someone hypes a stock as a "core supplier," you can ask the AI to verify that company's actual position in the supply chain and the strength of its customer evidence. What makes this project notable is its strict research boundary. It explicitly refuses to make buy or sell decisions, that stays with the user. Instead, it outputs prioritized research checklists, evidence chains, and next-step verification tasks. The repository includes a Python script for scoring individual companies against bottleneck criteria, along with example outputs showing what a finished research summary looks like. Everything is designed to turn vague market excitement into structured, evidence-backed investigation.