A repeatable research workflow

The Skill helps your AI choose the right tools, check dates, and turn observations into a clear research note. MCP provides the data; the Skill provides the workflow.

How the Skill works

Identify the ETF or stock and the observation period.
Use the smallest set of tools that answers the question. Prefer a single comparison or brief over repeated snapshots.
Check source dates, missing observations and differences between raw and scale-adjusted changes.
Present facts, uncertainty and source links in your language. Never turn missing data into zero or promise a return.

Install the Skill

Download the file and save it as active-etf-research/SKILL.md in a skill directory supported by your AI client. Connect the MCP server separately. A Skill file alone does not sign you in or grant data access.

For ChatGPT, use its supported Skill/plugin installation workflow if available on your account. Adding a custom MCP connector does not automatically install this Skill. You can still use the research prompts below with MCP alone.

For Grok Build, save the folder under ~/.grok/skills/. The Grok chat website uses the MCP connector; do not assume it imports local Skill files.

Claude

Claude Code: paste the one-prompt install request into Claude Code, then approve OAuth in your browser. Claude web: add this MCP URL under Customize → Connectors and enable it in the chat; local Skill files are separate.

Save the downloaded SKILL.md at ~/.claude/skills/active-etf-research/SKILL.md. Start a new Claude Code session to load it. The Skill file alone does not grant access.

The web connector does not install your local SKILL.md. Use the research prompts below; if your Claude account offers Skills, import the Skill separately.

Download SKILL.md

Try asking

Compare 00981A and 00991A: shared holdings, different exposures, and source dates.
Summarize the last 7 days of holding changes for 00981A. Separate raw changes from scale-adjusted changes.
Which tracked ETFs hold 2330? Show their source dates and recent observed changes.
Compare US-listed ETFs DRAM and NASA: shared holdings, different exposures, and each issuer's data date.

What this service does not do

No trading, personalized buy/sell instructions or advertising. A data date is not a real-time quote. Company and fund names remain in their source language.

Covers supported Taiwan-listed and US-listed ETFs. Compare funds within one listing market; issuer data dates can differ. 13F portfolios are not included.