Skip to content
Get packs

Guide · Agent skill

How to use the GEO Schema structured data auditor skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

The geo-schema skill is an agent skill for Claude that audits a web page's structured data, which is the hidden machine-readable markup (such as Organization, Article, Product or FAQPage) that tells search and AI tools what a page describes. It is useful for website owners, SEO staff, developers and publishers who want their markup checked and extended. After this guide you will know what the skill does, how to install it, how to ask for an audit in plain language, and what to check before you publish any generated code.

GEO Schema & Structured Data Auditorzubair-trabzada/geo-seo-claude · MIT

Quick start

  1. 1Download the ZIP. Download the geo-schema skill ZIP from this page.
  2. 2Unzip into a skills folder. Unzip it into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder.
  3. 3Or upload in the Claude apps. Alternatively, go to Settings → Capabilities → Skills in the Claude apps and upload the skill.
  4. 4Install the fetch script. Make sure Python 3 is installed and fetch_page.py exists at ~/.claude/skills/geo/scripts/, which comes from the separate geo skill.
  5. 5Ask for an audit. Name the page URL and describe what you want, such as checking Organization or Article markup.
  6. 6Review GEO-SCHEMA-REPORT.md. Read the report for missing, invalid or unverified items before changing anything.
  7. 7Verify and deploy the code. Check every URL and fact in the generated JSON-LD, then have someone with site access paste it in.

What the geo-schema skill makes the agent do

An agent skill is a folder of instructions that Claude can load when a task matches it. The geo-schema skill tells the agent how to run a structured data audit of a live page. It describes the source as a Schema.org audit and generation tool that detects, validates and generates JSON-LD markup.

The workflow in the skill has six steps: fetch the page HTML, detect existing structured data (JSON-LD, Microdata and RDFa), validate each block against Schema.org, identify recommended schemas missing for the business type, generate ready-to-use JSON-LD code blocks, and write a GEO-SCHEMA-REPORT.md file.

  • Detects JSON-LD, Microdata and RDFa on a page
  • Checks JSON-LD syntax, @type, required and recommended properties, sameAs links, URL validity and nesting
  • Flags schema that only appears after JavaScript runs
  • Generates JSON-LD code blocks and a GEO-SCHEMA-REPORT.md summary

Install the geo-schema skill from the ZIP

Download the ZIP for the geo-schema skill from this page. Then install it in one of two ways: unzip it into your personal skills folder at ~/.claude/skills/, or unzip it into a project's .claude/skills/ folder so it applies only to that project.

You can also upload the skill in the Claude apps under Settings → Capabilities → Skills. Choose whichever fits how you use Claude. Personal installs work across your projects; project installs keep the skill tied to one codebase.

  • Personal install: unzip into ~/.claude/skills/
  • Project install: unzip into the project's .claude/skills/ folder
  • Claude apps: Settings → Capabilities → Skills, then upload the skill

Set up the required fetch script

The skill does not include everything it needs. Its detection step runs fetch_page.py, which is not in the file you download. The skill depends on a separate geo skill installation, so fetch_page.py must exist at ~/.claude/skills/geo/scripts/.

You also need Python 3 available on your machine, and network access so the script can retrieve the target URL. Without the script, the audit cannot start, so check this before you ask for a report.

  • Python 3 installed
  • fetch_page.py located at ~/.claude/skills/geo/scripts/ (from the separate geo skill, not included in this ZIP)
  • Network access to the URL you want audited
python3 ~/.claude/skills/geo/scripts/fetch_page.py <url> page

Trigger the skill with a natural request

You do not need a special command. Describe the task in plain language and name the page. The agent should pick the skill when the request is about structured data, schema markup, JSON-LD or AI discoverability of a site.

Be specific about the goal. A request that names the page, the business type and the output you want gets a more useful result than a general question.

  • Audit the structured data on https://example.com/about and tell me what is missing
  • Check the Article markup on this post, including author and dateModified
  • Write Organization JSON-LD for my company and list any sameAs links I still need to add
  • Review whether my product page markup is valid and complete for Product schema

Read the report and the generated code

The audit output includes a structured_data array of parsed JSON-LD blocks from the page, and the skill writes GEO-SCHEMA-REPORT.md. Read the report first: it lists what exists, what fails validation, what is missing for the business type, and what rendering method the markup uses.

Any generated JSON-LD is a draft. Compare every name, URL, logo, address and sameAs link against facts you know are true before pasting it into your site.

  • Check each URL resolves and points to a page you control
  • Remove any sameAs link you cannot confirm
  • Confirm that facts such as founding date, address and job title are accurate
  • Deploy the markup in server-rendered HTML where possible

Safety notes: scripts and network access

This skill runs scripts and uses the network. Its allowed tools include Bash, WebFetch and Write, and it fetches the page you name with fetch_page.py. Only audit URLs you are comfortable having requested, and review commands the agent proposes before approving them.

The skill can write files, including GEO-SCHEMA-REPORT.md. Check where output is saved in your project before committing it.

  • Runs scripts: yes
  • Uses network: yes
  • Only use it on sites you own or manage, since someone with access must review and deploy the changes

Tips and limits

Use the skill for accuracy and completeness, not as a promise of results. The skill's claims that schema raises citation probability are not backed by evidence in the source file, and it cannot guarantee citations or rankings.

WebFetch strips head content where JSON-LD usually sits, so the skill uses fetch_page.py for detection. Prefer JSON-LD over Microdata or RDFa, which the skill flags for migration because they are harder for crawlers to parse.

  • Start with an Organization block on a business site, then add Article and Author on publishing pages
  • For LocalBusiness, add address, telephone and openingHoursSpecification
  • Do not add FAQPage expecting rich results; the skill notes these are restricted to government and health sites, though the markup can still serve AI readability
  • Re-audit after each deployment

What people use it for

  • Auditing an existing business homepage for Organization markup and missing sameAs links
  • Checking author and dateModified markup on blog posts for a content publisher
  • Generating JSON-LD snippets for a LocalBusiness contact or location page
  • Reviewing Product markup on an e-commerce page for required offers and brand fields
  • Reviewing whether a SaaS page has SoftwareApplication or WebSite with SearchAction markup

Frequently asked questions

How do I install the geo-schema skill?

Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. You can also upload it in the Claude apps under Settings → Capabilities → Skills.

What do I need to run the geo-schema skill?

You need Python 3, network access to the target URL, and a fetch_page.py script at ~/.claude/skills/geo/scripts/. That script is not in the ZIP and comes from a separate geo skill installation.

Is the geo-schema skill free?

The skill is released under the MIT licence, as stated on this page. Check the licence file in the source repository for full terms before reuse.

Does this skill guarantee more AI citations?

No. The skill cannot guarantee citations or rankings, and the claims about citation probability in its source are not backed by evidence in the file. Treat it as a way to check and improve markup accuracy.

Why does the skill use fetch_page.py instead of WebFetch?

WebFetch converts HTML to markdown and strips the head content where JSON-LD blocks usually sit, so it would miss them. fetch_page.py returns a structured_data array of parsed JSON-LD blocks.

Can I paste the generated JSON-LD straight into my site?

Not without checking it first. Generated markup can contain invented sameAs links or facts, which would mislead search engines, so verify each value and have someone with access deploy it.

How does geo-schema compare with a normal schema validator?

The skill combines detection, validation, gap analysis and code generation in one agent workflow, and it writes a GEO-SCHEMA-REPORT.md summary. A standalone validator usually checks only the markup you give it, so you would still need to write the missing code yourself.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Use the geo-schema skill if you are a website owner, SEO specialist, developer or publisher who controls a live site and wants its structured data audited and drafted quickly. Skip it if the site is not yet live, you cannot deploy changes, or you expect a guaranteed boost in AI citations, since the output still needs human review and the skill cannot promise results.

Download the skill