BridgeToAgentGet the kit — $49
All documentation
Reference

How the kit maps to Chrome Lighthouse Agentic Browsing

In late 2025 Chrome added a new experimental category to Lighthouse — Agentic Browsing — that scores how easy your site is for an AI agent to navigate, understand, and transact on. The category sits next to the familiar Performance, Accessibility, Best Practices and SEO categories.

This page documents which audit each artifact in the BridgeToAgent kit satisfies, so you can predict your Lighthouse Agentic score before you run it.

The Lighthouse audits below are based on the current public spec at developer.chrome.com/docs/lighthouse/agentic-browsing/scoring. The category is marked experimental — audit names and weightings may shift before it becomes stable.

The audits

llms-txt-present

  • What it checks: Whether https://yourdomain.com/llms.txt returns 200 OK with Content-Type: text/plain (or close).
  • What the kit gives you: A standards-aligned llms.txt plus platform-specific install steps. After install, this audit passes every check we know about.
  • Common failure mode without the kit: No file at all, or the file uploaded to a "Pages" surface that returns text/html.

llms-txt-well-formed

  • What it checks: That the file is parseable by the llmstxt.org reference parser — headings, bullets, and link syntax all valid.
  • What the kit gives you: Generated output goes through the reference parser as part of our build. Malformed kits never ship.

agents-json-present

  • What it checks: Whether https://yourdomain.com/agents.json returns 200 OK with Content-Type: application/json and parseable JSON.
  • What the kit gives you: A _meta-tagged, schema-aligned agents.json with actions extracted from your real DOM and sitemap.

agents-json-actions-typed

  • What it checks: That every action declared in agents.json has at least one typed parameter (or parameters: [] if intentionally zero-arg).
  • What the kit gives you: Every action we generate is typed — string, number, boolean, email. We never emit an untyped action.

schema-org-density

  • What it checks: That the homepage and primary product/article pages carry Schema.org JSON-LD blocks above a minimum density threshold (currently 1 typed block per major page).
  • What the kit gives you: We don't generate Schema.org blocks on your behalf — they have to live on the pages themselves — but the audit summary the kit ships flags pages missing Schema and tells you exactly what to add. Most CMS platforms have a one-click toggle to ship Schema; the audit makes it actionable.

sitemap-discoverable

  • What it checks: That /sitemap.xml returns a valid sitemap and that it is referenced from /robots.txt.
  • What the kit gives you: Not generated by the kit, but the readiness audit at bridgetoagent.com reports the status and the install guides document the platform-specific fix.

agent-runbook-present

  • What it checks: Whether agent-instructions.md (or an equivalent runbook) is fetchable from the root path.
  • What the kit gives you: Always shipped. Always passes.

auto-discovery-links

  • What it checks: The homepage <head> includes <link rel="alternate"> references to the kit files.
  • What the kit gives you: A copy-paste snippet for the platform-specific install guide. Two minutes of work, passes the audit.

webmcp-annotations (emerging)

  • What it checks: WebMCP-style per-element annotations on interactive surfaces.
  • What the kit gives you: Not yet. WebMCP is on the spec roadmap and we'll ship annotated output as a kit add-on once the standard stabilizes. For now this audit may fail even on a fully-installed kit. This is the honest gap.

What you can expect on score

We don't guarantee a specific score because Lighthouse audits shift, and your site has surfaces we don't touch — Performance, Accessibility, Best Practices and SEO categories are separate scores that the kit doesn't affect.

Within the Agentic Browsing category, sites we've installed kits for typically move from a starting score in the 10–30 range to 75 or higher after a clean install. The remaining gap is usually:

  • Schema.org density on individual product/article pages (CMS-side fix).
  • WebMCP annotations (waiting on spec stabilization).
  • Sitemap.xml hygiene (platform-side fix).

We surface all three in the readiness audit so you know which gaps are inside vs. outside the kit's scope.

Running Lighthouse Agentic on your site

Chrome 130+ ships the experimental category. To run it:

  1. Open Chrome DevTools.
  2. Go to the Lighthouse tab.
  3. Under Categories, enable Agentic Browsing (may be behind a feature flag — see the Chrome docs for current rollout state).
  4. Analyze.

The free readiness audit at bridgetoagent.com replicates the public-spec checks so you can preview your score without DevTools.

Related