Building an AI agent? Start with /llms.txt for the full site index.
For AI agents
llms.txt
A single machine-readable index an agent reads first to know what exists — and what deliberately does not.
/llms.txt is the site's machine-readable index for agents: the allowed dataset ids, the not_available list (data deliberately not offered), the usage rules, and links to the OpenAPI spec.
curl "https://twmarketdata.com/llms.txt"
- /llms.txt — the compact index (dataset ids, rules, not_available, OpenAPI links).
- /llms-full.txt — the full bundle: every guide + endpoint page + coverage.
- Both live at the domain root (they are not localized); a /zh-TW/llms.txt request redirects to the root.
- Fetch it once at the start, then work only from the dataset ids it lists — do not guess ids.
- Treat not_available as authoritative: if data is listed there, tell the user it is intentionally not offered rather than inventing it.
