Source-led article
Lighthouse Update Fails llms.txt Files Without Markdown Links

A recent update to Google Lighthouse, version 13.3.0, introduces an Agentic Browsing audit category that is causing `llms.txt` files to fail if they do not adhere to Markdown link syntax. This development is significant for web developers and SEO professionals in India, as it impacts how AI agents are expected to interpret website instructions and could affect site visibility and performance audits.
The new audit interprets `llms.txt` as a Markdown document, even when served as plain text with a `.txt` extension. This means that traditional plain-text links within the `llms.txt` file, which might be perfectly functional for human readers or other parsers, are now flagged as invalid by Lighthouse. The fix, as demonstrated by early adopters, often involves a simple conversion to Markdown link format.
Key facts
| Audit Name | Lighthouse Version | Requirement | Impact |
|---|---|---|---|
| llms.txt discoverability | 3.0 | Markdown link syntax | Audit failure if not followed |
| Agentic Browsing | 3.0 | New category | Fractional pass ratio |
The Problem with Plain Text
The core issue lies in Lighthouse’s strict interpretation of the `llms.txt` specification, which defines the format as a Markdown document. While many websites currently use a more natural plain-text format for their `llms.txt` files—listing links with simple descriptions—Lighthouse’s parser demands that every link be encoded using the `text` Markdown syntax. If this specific format is not used, the audit reports “File does not appear to contain any links,” leading to a failure.
For instance, a line like “- Homepage: /” would fail, whereas “- Homepage” would pass. This seemingly minor discrepancy highlights a crucial shift in how AI agent instructions are expected to be structured for optimal discoverability and parsing by tools like Lighthouse. The audit’s strictness means that even accurate and comprehensive `llms.txt` files can fail if they do not conform to this specific syntax.
Implications for Indian Websites
This change has immediate implications for websites across India, particularly those relying on `llms.txt` for AI agent guidance. Developers and SEO specialists will need to review their existing `llms.txt` implementations to ensure compliance. Failure to update these files could result in lower Lighthouse scores, potentially affecting perceived site quality and how AI agents interact with the content.
Many popular WordPress plugins, such as AIOSEO, already generate `llms.txt` files with Markdown link syntax by default. This suggests that websites using such plugins might be less affected. However, sites with hand-curated or older `llms.txt` files are highly likely to encounter this audit failure. The quick fix usually involves converting existing links to the `text` format and replacing separators like hyphens with colons if used for descriptions.
Understanding the Agentic Browsing Category
The Agentic Browsing category itself is new in Lighthouse 13.3.0 and includes six audits, such as `accessibility-tree well-formedness`, `cumulative-layout-shift`, and `llms.txt discoverability`. Unlike other Lighthouse categories that provide a 0-100 score, Agentic Browsing returns a fractional pass ratio, reflecting that the standards for the “agentic web” are still evolving. This indicates that Google is actively developing how AI agents will interpret and navigate websites, making adherence to these emerging standards increasingly important.
The audit’s design aims to ensure that `llms.txt` files are mechanically parseable by AI agents. While a human-readable plain-text file might convey the same information, programmatic extraction of links requires a structured format. Markdown link syntax provides this structure, making it easier for AI agents to identify and utilize the provided navigation paths and resources.
Next Steps for Developers
Given this update, Indian webmasters and developers should:
1. Audit current `llms.txt` files: Check if your `llms.txt` uses Markdown link syntax `text`.
2. Update non-compliant files: Convert any plain-text links to the required Markdown format.
3. Re-run Lighthouse audits: Verify that the `llms.txt discoverability` audit now passes.
4. Stay informed: Keep an eye on further updates to Lighthouse and the `llms.txt` specification as the agentic web continues to develop.
Source: Search Engine Journal (https://www.searchenginejournal.com/lighthouse-fails-your-llms-txt-without-markdown-links/577590/)