agentleFS
peakanswerpeakanswer/peak-answer-skills/skills/llms-txt/SKILL.md

llms-txt

Skill0 starsChanged 49 days agoMITView on GitHub

Write or review an llms.txt file for a site. Use when someone asks what llms.txt is, wants one written, wants theirs checked, or asks whether it is worth having.

14 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use content-brief.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • LangChainSkill

    content-brief

    Turn a vague or multi-asset design ask into a brief. Pull what is already known, ask at most two questions with reasons, decide the rest and say what you decided.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 14 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not describe it as required, as a ranking factor, or as something engines expect.

    line 14
  2. Distinctive:

    Do the links have descriptions, and are they absolute?

    line 58
  3. Distinctive:

    Do the links resolve? A dead link here is worse than a missing one.

    line 59

“Common” is measured: kept by at least 6 in 10 of the 14 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

llms.txt

A markdown file at https://<domain>/llms.txt that tells a model what a site is and points it at the pages worth reading. It is a proposal, not a standard, and no engine has committed to reading it.

Say the honest thing first

Adoption is contested. No major engine has confirmed it uses llms.txt for retrieval or ranking, and there is no measured evidence that adding one lifts citations. It costs an hour and it cannot hurt, which is the whole case for it. If someone is choosing between writing llms.txt and fixing a Disallow: / on GPTBot, the robots file wins by a distance.

Do not describe it as required, as a ranking factor, or as something engines expect. That claim is everywhere and it is not true.

The format

Plain markdown, no front matter. One # heading, a > blockquote summary, then ## sections of links with a short description after each.

# Acme Analytics

> Session replay and funnel analysis for B2B SaaS teams. Self-serve from $49 a month, SOC 2 Type II, EU data residency available.

Acme records what users do in a product and shows where they drop out. Used by product and growth teams, typically 10 to 200 seats.

## Docs

- [Quickstart](https://acme.com/docs/quickstart): install the snippet and see a first session in about five minutes.
- [Funnels](https://acme.com/docs/funnels): defining a funnel, and how drop-off is attributed between steps.
- [Data retention](https://acme.com/docs/retention): what is stored, for how long, and how to shorten it.

## Product

- [Pricing](https://acme.com/pricing): plans, seat counts and what each tier includes.
- [Security](https://acme.com/security): SOC 2, subprocessors, data residency.

## Optional

- [Changelog](https://acme.com/changelog): releases, newest first.

Rules that matter:

  • Absolute URLs. The file may be read far from its origin.
  • A description after every link, separated by a colon. A bare list of URLs is the most common failure and it carries almost no information.
  • ## Optional is special. Its contents are what a model can skip when short of context. Everything else is what you want read.
  • Keep it under about 50 links. A dump of the sitemap is a sitemap, not this.

What goes in the summary

The blockquote is the piece most likely to be read in isolation. It should answer what this is, who it is for, and the two or three facts a buyer needs: price shape, deployment model, compliance, region. Specifics, not adjectives. "Leading platform for modern teams" tells a model nothing it can quote.

Reviewing an existing one

  1. Does it exist and return 200 as text/markdown or text/plain, not an HTML 404 page?
  2. Is there a single # heading and a > summary under it?
  3. Do the links have descriptions, and are they absolute?
  4. Do the links resolve? A dead link here is worse than a missing one.
  5. Is it current? A file listing last year's product names dates the site to whatever reads it.
  6. Is it consistent with the site itself? Contradicting the pricing page is worse than having no file.

llms-full.txt

The same idea with the page contents inlined rather than linked. Only worth generating if the docs are small and genuinely stable, because it goes stale silently and nothing tells you.

With the Peak Answer MCP connected

generate with llms_txt writes one from the site's own pages, in this format, with the descriptions drawn from the pages rather than invented. Read it before publishing: it is a draft that knows the site's structure, not the positioning.

Free, no signup: https://peakanswer.com/tools/llms-txt-generator

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.