[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"glossary-llms-txt::en":3,"gloss-cluster-llms-txt::en":23,"gloss-next-llms-txt::en":9},{"slug":4,"category":5,"name":6,"definition":7,"meta_desc":8,"faq":9,"schema_markup":9,"related":10},"llms-txt","dev-tools","llms.txt","llms.txt is a proposed convention: a Markdown file served at the root of a domain that lists the pages a language model should read to understand the site, each with a one-line description, and optionally links to clean Markdown versions of those pages. It borrows its placement and its spirit from robots.txt, but it is an invitation rather than a restriction — a curated reading list instead of an exclusion list.\n\nThe problem it responds to is real enough. A model reading a marketing site through raw HTML spends its limited context on navigation, consent banners, script tags and boilerplate before reaching a sentence worth having. A retrieval client landing on a documentation site has no way to know which twenty of two thousand URLs actually matter. A short, human-curated index is a cheap way to say so.\n\nIt is a convention and not a standard, and the difference matters. It is not part of the robots protocol, no search engine treats it as a directive, and a publisher cannot verify from outside whether anything reads the file at all. Publish it because the curation is useful in its own right — it doubles as an orientation page for humans and for agents you build yourself — and not on the assumption that a file is consumed because it exists.\n\nWhat separates a good one from a decorative one is maintenance. It should be short, honest, and generated from the same source as your navigation or sitemap wherever possible. A list that describes the site as it stood on launch day is worse than no list, because whatever is reading it has no way to tell that it has gone stale and will quote the description rather than the page.\n\nIt is not access control and not a security boundary. List only URLs that are already public, put nothing in it you would be unhappy to see quoted verbatim, and do not use it to try to hide anything — a file at a predictable path is the first thing anyone curious will fetch.","llms.txt is a proposed Markdown index at a domain's root telling language models which pages matter. A convention, not a standard — publish it, expect nothing.",null,[11,14,17,20],{"slug":12,"name":13},"cms","Content Management System (CMS)",{"slug":15,"name":16},"grounding","Grounding",{"slug":18,"name":19},"retrieval","Retrieval",{"slug":21,"name":22},"static-site-generator","Static Site Generator (SSG)",[24,28,32,35,38,41,44,47,50,53,56,59],{"slug":25,"category":5,"name":26,"updated_at":27},"abstraction-layer","Abstraction Layer","2026-09-08T21:32:37+00:00",{"slug":29,"category":5,"name":30,"updated_at":31},"agent","Agent","2026-09-08T21:32:35+00:00",{"slug":33,"category":5,"name":34,"updated_at":31},"ai-code-assistant","AI Coding Assistant",{"slug":36,"category":5,"name":37,"updated_at":31},"api-gateway","API Gateway",{"slug":39,"category":5,"name":40,"updated_at":31},"api-versioning","API Versioning",{"slug":42,"category":5,"name":43,"updated_at":31},"autonomous-agent","Autonomous Agent",{"slug":45,"category":5,"name":46,"updated_at":31},"blue-green-deployment","Blue-Green Deployment",{"slug":48,"category":5,"name":49,"updated_at":31},"canary-deployment","Canary Deployment",{"slug":51,"category":5,"name":52,"updated_at":27},"canonical-url","Canonical URL",{"slug":54,"category":5,"name":55,"updated_at":31},"chaos-engineering","Chaos Engineering",{"slug":57,"category":5,"name":58,"updated_at":31},"ci-cd","Continuous Integration \u002F Continuous Deployment (CI\u002FCD)",{"slug":60,"category":5,"name":61,"updated_at":62},"circuit-breaker","Circuit Breaker","2026-09-08T21:32:36+00:00"]