Skip to content

Fix structured data and server rendering

Who it is for

A developer who owns a site’s code and wants an agent to fix what an AEO audit flags.

The job

Turn a list of failing rules into code changes in your stack, then prove each change landed.

Tools and skills

  • analyze_website_aeo on the aeo-research server. It crawls 10 pages and returns at most the top 8 fixes.
  • The Claude Code plugin, which installs 24 skills and both servers: /plugin install aeo-toolkit --marketplace Advance-Labs/aeo-toolkit.
  • Every audit rule has exactly one owning skill. The ones you will use most for code:
  • The rule definitions are listed under audit rules and AEO rules.

Steps

  1. Install the plugin and open your repo in Claude Code.
  2. Ask the agent to run the aeo skill on your site. It calls analyze_website_aeo, groups the failures and orders them by impact.
  3. For each group, the owning skill follows the same loop: detect with the rule, decide whether it is a real problem, implement the recipe for your framework, then verify with the same check.
  4. For rendering, test what the server sends rather than what the browser shows. Fetch the page with curl and look for your content in the raw HTML. If it is missing, the skill covers SSR, static generation and prerendering for your framework.
  5. For structured data, validate that each JSON-LD block parses and that the visible page matches the markup. The aeo-structured-data-validation skill covers duplicates from plugins and colliding @id values.
  6. Deploy, then run analyze_website_aeo again. The failing rule should now pass.

What you get

Code changes in your repo, one group at a time, and a re-run of the same check as proof.

The audit that finds these failures runs this pipeline. The extension skips the crawl and reads the rendered page.

The audit pipeline: the crawler fetches pages, the parser reads them, the schema validator checks JSON-LD, scoring runs 44 rules, and the report lists the fixes.

Limits

  • analyze_website_aeo is a shortlist. A rule that is not in the top 8 can still fail. Use the hosted audit for the full list.
  • A passing audit does not mean AI will cite you. Measure that separately, as in Perplexity citations vs competitors.
  • The agent edits your code. Review each diff.

AEO Toolkit is an open-source instrument by Advance Labs.