Tutorials

Lighthouse llms.txt Audit: Server Errors, Optional Files & Local Checks

Understand the experimental Lighthouse audit, distinguish a server error from an optional missing file, and review llms.txt text without treating a local score as a Chrome result.

By Muhammad Saqlain
·Published 2026-09-28·Updated 2026-10-20·5 min read

Use this guide with

2 ToolsWebPro tools

Open the free tool(s) below and follow the steps in this guide.

First identify the actual audit result

An llms.txt text-format review and a Lighthouse network audit answer different questions. A local score cannot tell you whether a public URL returns a server error, redirects to a login page or serves the expected document.

According to Chrome’s llms.txt audit documentation, an absent file returning 404 is Not Applicable, while a server error retrieving it is flagged. Use the actual report and response when troubleshooting. The audit is experimental.

Check the deployed HTTP response separately. Do not create a file solely to resolve a supposed mandatory 404 syntax failure.

Separate proposal structure from local preferences

The current llms.txt proposal requires an H1 project or site name. A summary and link sections are optional. If used, they should explain the site clearly and point to useful, accurate documentation.

ToolsWebPro also offers local review notes about brevity, relative URLs and a 50,000-byte target. Those are its own guidance, not official Lighthouse error codes or universal consumer limits. Absolute URLs can make copied text clearer, but URL rewriting needs the correct base.

↔ Scroll table horizontally to view full data
CheckWhat it establishesWhat it does not establish
Project-name H1A local structural checkA deployed Chrome audit result
Summary, sections and size notesOptional review preferencesA mandatory Lighthouse syntax checklist
Basic link syntaxSome text-format problemsThat the destination exists or is accessible
Local scoreA heuristic for this checkerRanking, AI citations or an official Chrome score

Use the generator without losing review control

  • Open the llms.txt tool and choose a template, or paste sitemap XML to import URL candidates. This does not crawl the linked pages.
  • Write a real project title and optional summary. Remove template claims, private URLs, tokens and information that should not be published.
  • Inspect each URL, its title and its relevance. A generated title inferred from a URL is not a verified page description.
  • Open Validate llms.txt and review the text findings. Apply suggested fixes only after checking the base URL; fixes can add placeholder text or alter URL schemes.
  • Copy or download the draft. The llms-full.txt view is an expanded outline in this interface; it does not fetch complete page bodies.

Check the actual published resource

For a root-level static file in Next.js, public/llms.txt is served as /llms.txt. Verify the deployed URL, expected response body and readable content type after your approved release. Deployment paths and cache behavior depend on the host; do not infer production behavior from a downloaded file.

Run the current Lighthouse workflow separately if you need its result. Chrome’s agentic scoring documentation describes an experimental category with a pass ratio rather than this tool’s 0–100 heuristic.

  • Open the public resource without authentication and inspect its response.
  • If it returns a server error, investigate the route, hosting and origin logs available to the maintainer.
  • Check destination links and remove obsolete or restricted resources.
  • Retest after deployment and record the version and date. A successful local check is not a production verification.

Keep AI-search expectations grounded

Google’s AI Search guidance does not require a special AI text file. Normal search eligibility and useful accessible content still matter. Other consumers may use different conventions, so test actual behavior rather than promising citations.

An llms.txt file is not an access-control mechanism. Review crawler preferences separately with the robots.txt generator and the crawler-control guide. A robots directive also does not replace authentication for private data.

Frequently asked questions

Does a missing llms.txt file fail the current Lighthouse audit?

Chrome’s experimental audit documentation treats an absent file returning 404 as Not Applicable. A server error retrieving the resource is flagged. Inspect the exact audit result and HTTP response rather than assuming a missing file is a syntax failure.

Which llms.txt sections does the proposal require?

The current proposal makes the H1 project or site name its only required section. A summary, background information and H2 link lists are optional. A local checker can offer additional preferences, which should be labelled as guidance.

Does the ToolsWebPro score come from Lighthouse?

No. It is a local text-review heuristic. It does not run Chrome, fetch the deployed file or establish server status, link availability, indexing or AI citations.

Will llms-full.txt contain the full website automatically?

No. The generator interface creates an outline from entered links and summaries. It does not fetch linked page bodies. Review the actual output and supply document text separately if a full-text reference is required.

Does Google require an AI text file for AI Overviews or AI Mode?

Google’s Search guidance says no special AI text file is required. The usual eligibility, indexing and content-quality requirements still apply; publishing an llms.txt file does not guarantee inclusion.

Free local text checker

Review your llms.txt text

Build an index, inspect local checks and review suggested edits before testing the deployed URL separately.

M

Muhammad Saqlain

Cybersecurity Practitioner & Lead Engineer

Security researcher, web developer, and founder of ToolsWebPro. Tests password entropy, GPU cracking speeds, and client-side encryption systems.

Read full author bio & credentials →