Blog > LLMO > llms.txt for Developer Tools: What It Does and Doesn’t Do llms.txt for Developer Tools: What It Does and Doesn’t Do Kirill SajaevSEO & Founder Nov 4, 2026 · 3 min read SHARE ON Table of contentsDoes llms.txt help developer tools?What llms.txt isThe formatExample for an API productWhat it helps: agentsWhat it doesn't help: AI search citationsChecklistRelated reading Docs series (part of GEO for PLG SaaS): Structuring docs for AI retrieval · llms.txt for developer tools · API references, code samples and changelogs · Coding agents and your dev tool Does llms.txt help developer tools? Yes, for coding agents; no, for AI search citations. An llms.txt file is a Markdown index of your docs that helps agents find the right page and read it cheaply. It doesn’t make AI search engines cite you more. Publish one, because it’s cheap and agents use it, but don’t count it as a GEO strategy. Your docs structure matters far more. See structuring docs for AI retrieval. What llms.txt is llms.txt is a proposal by Jeremy Howard, first published in September 2024 and revised as v2 in August 2026. It defines a Markdown file at /llms.txt, or at a subpath such as /docs/llms.txt, that gives agents a short summary of a site and links to LLM-friendly versions of its key pages. The spec says a file covers the URLs under its path, and agents should use the most specific one. It also proposes that pages offer a clean Markdown version at the same URL with .md added or swapped in. v2 adds standard ways to find those: rel="alternate" type="text/markdown" for a page’s Markdown version and rel="describedby" for the llms.txt that covers it. The format The spec defines these sections, in this order: An H1 with the project or site name. This is the only required part. A blockquote with a short summary. Optional paragraphs or lists with more detail (no headings). Optional H2 sections, each a list of links in the form [name](url): notes. An “Optional” H2 section is a convention for secondary links an agent can skip. Example for an API product # Acme Payments API > Acme Payments is a REST API and TypeScript/Python SDKs for accepting card and bank payments. Current API version: 2026-10. Base URL: https://api.acme.example All endpoints need an API key in the Authorization header. Amounts are integers in the smallest currency unit. ## Get started - [Install the SDK](https://docs.acme.example/install.md): npm and pip install commands - [Make your first payment](https://docs.acme.example/first-payment.md): end-to-end example in TypeScript and Python - [Authentication](https://docs.acme.example/auth.md): API keys, test vs live mode ## API reference - [Payments](https://docs.acme.example/reference/payments.md): create, capture, refund - [Webhooks](https://docs.acme.example/reference/webhooks.md): event types and signature checks - [Errors](https://docs.acme.example/reference/errors.md): every error code and fix ## Changes - [Changelog](https://docs.acme.example/changelog.md): dated, versioned changes - [Migrate from v2025-04](https://docs.acme.example/migrate-2025-04.md): breaking changes and code diffs ## Optional - [Concepts](https://docs.acme.example/concepts.md): payment lifecycle overview Good files are short, list the pages an agent needs to write working code, state the current version and point to Markdown URLs. Real examples: Claude Code’s docs and the Model Context Protocol docs both publish one. What it helps: agents Coding agents and IDE assistants can read the file to find the right page instead of crawling your whole site, then fetch Markdown with less noise. Docs MCP servers and tools can use it as the index for what to serve. Your own instructions. You can point a coding agent at it from a project instruction file. See getting your dev tool used by coding agents. Agents don’t always look for it on their own. Feature requests on Claude Code’s GitHub have asked for automatic llms.txt fetching (#43728, #2476), and a commenter on #2476 suggests the most reliable workaround is to tell the agent about it in CLAUDE.md. What it doesn’t help: AI search citations AI search engines cite HTML pages, not Markdown files. Surferstack reports Promptwatch data showing HTML accounted for 99.94% of citations in a seven-day sample, with Markdown files at about 0.05%. So llms.txt won’t: get you onto “best [category] tools” shortlists replace server-rendered, well-structured HTML docs fix wrong facts AI answers repeat about your product Checklist [ ] /llms.txt (or /docs/llms.txt) exists, starts with an H1 and a one-line summary. [ ] It names the current version and base URL. [ ] It links to install, quickstart, auth, reference, errors and changelog pages, and to a pricing page that states the free plan or trial plainly. [ ] Links point to Markdown versions that return clean text. [ ] It’s generated from your docs build, so it can’t go stale. [ ] Your HTML docs are still server-rendered for search and citations. Related reading GEO agency for dev tools How AI recommends developer tools What is GEO? Best GEO agencies for developer tools in 2026 What is AI share of voice, and how is it measured? Want to know whether AI answers recommend your dev tool? Check your AI visibility for free. Book a Free SEO consult We love chatting SaaS SEO and we've been doing it for a long time, book a call and walk away with a free SEO strategy! Kirill SajaevSEO & Founder Get a Free consult
LLMO How to get high‑quality developer content from a small marketing team How we did it at DeepInfra and LlamaIndex Marketing to developers comes with its own set of challenges The audience is smart, detail-oriented and ...
LLMO The content strategy that makes AI agents position you accurately Cut through the noise and get more signups Most dev-tools companies are described five different ways: by their homepage, their docs, their README, ...