what is llms-txt-validator fr?

This tool checks whether a website's llms.txt file, a plain text page meant to describe a site to AI agents, is structured correctly. It runs seven specific checks against the file and reports each one as pass, warn, or fail, along with a short explanation, rather than giving a single overall score, since the author felt a numeric score would imply a measurement of agent readiness that these structural checks do not actually make. The checks look at things like whether the file exists and returns a successful response, whether it is plain text rather than HTML, whether it starts with a proper title, whether it includes a short one-line summary, whether it is organized into clear sections, whether its links are ones an AI agent could actually follow, and whether the file is small enough to be cheap to read. An extra check flags if the file contains raw HTML mixed in with the text. You can use it as a command line tool by installing it globally or running it directly with npx against any domain, get the results as JSON for automated processing, or use a strict mode where warnings also count as failures, useful for plugging into a CI pipeline like GitHub Actions. It can also be used as a plain JavaScript function inside a Node project to check a site's llms.txt file programmatically. This project is the open-source version of the same checking logic that also runs as a hosted web tool at the author's own site, turva.dev, which the author says is treated as the authoritative version if the two ever disagree. The fetch itself only reads the target site's file over a secure connection, follows same-site redirects only, times out quickly, and does not store anything. The project is released under the MIT license.