A review of 438 `llms.txt` files found in public GitHub repositories reveals that while most adhere to the initial formatting of an H1 heading and a summary blockquote, a significant portion deviates from the proposed structure. Specifically, over half of the files include prose within sections designated for link lists under H2 headings, and a smaller percentage utilize deeper heading levels or multiple H1s. While these deviations do not prevent AI agents from reading the files, they reduce the predictability of link discovery, which is a key goal of the `llms.txt` format. AI
IMPACT Highlights potential challenges in standardizing AI agent data formats and predictable link discovery.
RANK_REASON Analysis of a specific file format's adherence to a proposal. [lever_c_demoted from research: ic=1 ai=0.4]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →