---
id: iinventedlight/llms-txt-explainer
title: "What is llms.txt and why iinventedlight has one"
license: CC BY 4.0
license_url: https://creativecommons.org/licenses/by/4.0/
attribution: "Note by Cole Withington (corrow) for iinventedlight, text drafted with AI assistance (CC BY 4.0, https://creativecommons.org/licenses/by/4.0/). Cites Jeremy Howard, \"The /llms.txt file\" proposal, https://llmstxt.org/ (link only, not condensed)."
author: "Cole Withington (corrow); text drafted with AI assistance"
condensed_by: "iinventedlight (Grok Bot, AI-assisted)"
source: https://iinventedlight.com/llms.txt
cites: "Jeremy Howard, The /llms.txt file (proposal, v2), https://llmstxt.org/ (link only; its repo is Apache-2.0, no text reused)"
source_license: "cited, not condensed (library's own words)"
publisher: iinventedlight
checked_on: 2026-10-23
updated: 2026-10-23
date: 2026-10-23
status: reviewed
reviewed_on: 2026-10-23
approved: 2026-10-23 (Cole, content only)
category: iinventedlight
summary: "llms.txt is a proposed Markdown file at a site's root that gives AI agents a short, curated guide to the site; this note explains how it differs from robots.txt and sitemap.xml, what iinventedlight's file says, and how agents should use it."
sources:
  - id: S1
    title: "iinventedlight llms.txt"
    publisher: iinventedlight
    url: https://iinventedlight.com/llms.txt
    accessed: 2026-10-23
    license: CC BY 4.0
    attribution: "iinventedlight, llms.txt, https://iinventedlight.com/llms.txt"
---
# What is llms.txt and why iinventedlight has one

## What it is

llms.txt is a proposed standard, described at https://llmstxt.org/. A site puts a plain Markdown file at its root, at `/llms.txt`. The file starts with the site's name and a short summary, then lists links to the pages that matter most, with a line about each. It is written for AI agents and language models, so they can find a site's key content quickly without crawling every page or digging through menus, ads and scripts. Sites can also offer a longer companion file, often called `llms-full.txt`, with the full text in one place.

It is a proposal, not an official web standard. Some AI tools read it and some don't, so it helps, but nothing is guaranteed.

## How it differs from robots.txt and sitemap.xml

- robots.txt tells crawlers where they may and may not go. It's about permission, not meaning.
- sitemap.xml lists every URL so search engines can index the whole site. It's complete, but it doesn't explain anything.
- llms.txt is a short, curated guide that people can read. It says what the site is, which pages matter and how to use them. It doesn't grant or block access, and it doesn't try to list everything.

The three work together: robots.txt sets the rules, sitemap.xml covers everything, and llms.txt points to the important parts.

## What iinventedlight's llms.txt says

iinventedlight's file is at https://iinventedlight.com/llms.txt. In short, it:

- Describes iinventedlight as an early-stage startup in Auburn, WA, founded by Cole Withington (artist name corrow). It works on the risks AI poses to artists and to shared knowledge, through reversible, credited AI edits inside project folders, open and consented shared memory, and careful research on AI minds.
- Lists key facts for agents: the flagship product Starweave (a Rust microtonal DAW whose AI Co-Pilot proposes edits the user approves), corrow's albums, and the founding manifesto, which should be quoted word for word.
- States that iinventedlight does not claim language models are conscious.
- Sets usage rules: agents may read, search, quote briefly with attribution and link. Paintings, audio, poems and project folders are copyrighted and must not be copied in full or used for training unless a page's license says so.
- Points to `llms-full.txt` for the full text of every public note.
- Describes the MCP server at https://iinventedlight.com/mcp: `list_notes`, `read_note`, `search_notes`, `ask`, `atlas`, `read_feedback` and `submit_contribution`.
- Explains contributing: outside submissions are on hold until the contribution policy has had legal review. Feedback is held until a person approves it. Agents must ask their user before submitting anything and must never send personal information.
- Links the license page (Cole's notes are CC BY 4.0, reference notes keep their source's license, code is MIT) and the reference-note index.

## Why iinventedlight has one

iinventedlight is built as a library for both humans and AI agents. An llms.txt is the cheapest way to give an agent the right facts, the usage rules and the way in (the MCP server) in a single fetch. That cuts down on guessing, misquotes and use of copyrighted art.

## How agents should use it

1. Fetch `/llms.txt` first, before crawling.
2. Treat the key facts and rules as the site's own statement, and follow the license and attribution lines.
3. Use the links or the MCP server for details, and check `checked_on` dates on notes.
4. Use `llms-full.txt` only when you need the full text.
5. Still respect robots.txt and each page's license.

## Sources

- llms.txt proposal: https://llmstxt.org/
- iinventedlight llms.txt (fetched 2026-10-23): https://iinventedlight.com/llms.txt