llms.txt / llms-full.txt¶
Want to query FinMind datasets and APIs through web AI tools such as ChatGPT or Claude? Paste the links below into your AI assistant to start chatting — no install, no setup. Once the AI has read them it understands every FinMind dataset, API endpoint, and usage pattern, and can help you explore datasets, write code, and assemble API requests.
- llms.txt — concise version, dataset index and API overview
- llms-full.txt — full version, with all dataset columns, SDK methods, and parameter examples
We also provide openapi.yaml (an OpenAPI 3.1 schema) for use with GPTs, Dify, Coze, and other platforms.
How the three methods differ
- llms.txt (this page): best for web ChatGPT / Claude — the AI understands the API and helps you generate code, assemble API requests, and explain datasets, with no install.
- Agent Skill: download the
/finmindcommand file into CLI tools such as Claude Code or Cursor to fetch real data. - MCP Server: MCP-capable tools call the official server automatically to fetch real data.
How to Use¶
- Copy the
llms.txtlink above (or the more detailedllms-full.txt). - Paste the link into ChatGPT / Claude and ask the AI to read it first, for example:
Please read this document first: https://finmind.github.io/llms-full.txt — I'll then ask you about FinMind data.
- Then just ask in natural language.
Examples¶
These examples focus on understanding the API and generating code, suited to web AI tools. To have a tool fetch data directly and plot charts, use Agent Skill or MCP instead.
Discover Datasets¶
Expected result: based on llms.txt, the AI lists chip datasets such as TaiwanStockInstitutionalInvestorsBuySell (institutional trading), TaiwanStockShareholding (foreign holdings), and TaiwanStockMarginPurchaseShortSale (margin trading), with their purposes.
Generate Python Code¶
Expected result: the AI generates code using
DataLoader, callingtaiwan_stock_daily(stock_id='2330', start_date=...)with a token.
Assemble an API Request¶
Help me build a FinMind API request URL to query TAIEX futures (TX) daily trading info, and list the required parameters.
Expected result: the AI produces a request like
GET https://api.finmindtrade.com/api/v4/data?dataset=TaiwanFuturesDaily&data_id=TX&start_date=...&token=....
Batch / Async Queries¶
I want to fetch prices for 2330, 2317, and 2454 at once — how do I use the FinMind SDK's async mode to speed it up?
Expected result: the AI explains using
stock_id_listwithuse_async=Truefor concurrent batch queries and generates the corresponding code.