HOME SKILLS BLOG GITHUB
// SKILL REFERENCE

FULL-SITE SEO AUDIT
FROM CRAWL TO ACTION

Render the homepage, crawl up to 500 reachable pages, coordinate specialist checks, and write a weighted audit report with a prioritized action plan. The result is a local diagnostic package, not a promise of rankings or indexation.

By Daniel Agrici · Page updated · Source verified against a reviewed Claude SEO source snapshot

$
/seo audit <url>

REQUIRES CLAUDE SEO INSTALLED IN CLAUDE CODE

A website passes through crawl, content, schema, performance, image, and AI-search checks into a prioritized audit report
// RUN PREVIEW

What will happen if you start this audit?

The command needs one reachable URL. It can run its core crawl without Google or DataForSEO credentials, fetches the public site, and writes a local evidence package without changing the audited website.

OUTCOME

A prioritized full-site diagnostic across the documented specialist categories.

REQUIRED INPUT

One reachable website URL supplied to /seo audit <url>.

READINESS

Core mode needs Claude SEO, its isolated runtime, and Chromium when rendering is required. Provider credentials are optional.

LOCAL OUTPUT

A domain folder with the report, action plan, JSON envelope, category findings, and available screenshots.

SIDE EFFECTS

Fetches the public site and writes local files. It does not edit the audited website.

COST + APPROVAL

No separate Claude SEO license fee. Claude Code access and optional provider calls can have separate costs and credential requirements.

// DECIDE

Should you use the full-site audit?

Use it for a site-wide baseline

Choose the audit when you need one prioritized view across crawl access, content, on-page SEO, structured data, performance, images, and AI-search readiness.

Do not use it for one isolated page

Use /seo page for a focused URL review. A full crawl adds time and broader findings that may not help a single-page decision.

What it can establish

It can record which reachable pages and specialist categories were checked, preserve evidence, and organize findings into a repeatable local artifact set.

What it cannot establish

Its health score is Claude SEO's prioritization model. It is not a Google score, ranking forecast, Core Web Vitals field pass, or guarantee that a URL will be indexed.

// WHAT HAPPENS

What happens when the command runs?

01
RENDER
The workflow renders the homepage in automatic mode and captures raw HTML, rendered HTML, extracted text, SPA status, and accessibility data when needed.
02
DETECT + CRAWL
It detects the business type from homepage signals, then follows internal links up to the 500-page ceiling while respecting robots.txt.
03
DELEGATE
Eight specialist roles always participate. Seven more are selected when the site type, credentials, stored baseline, or content signals make them relevant.
04
SCORE
Findings are aggregated into seven weighted categories and a 0 to 100 internal health score used to order the work.
05
PERSIST
The audit writes its report, action plan, structured JSON envelope, category findings, and available screenshots under one domain-specific folder.
06
REPORT
The result is a severity-ordered implementation plan. The reviewed command contract also recommends offering an optional PDF report after the audit completes.

Specialist routing

RouteSpecialistsCondition
Always-onTechnical, content, schema, sitemap, performance, visual, GEO, SXOIncluded in every documented full audit
ConditionalLocal, maps, Google, backlinks, cluster, drift, ecommerceSelected from site signals, credentials, APIs, or an existing baseline

Optional extensions can enrich these findings. For example, Google credentials can add CrUX, Search Console, and Analytics data, while DataForSEO can add live search, backlink, on-page, listing, and AI-visibility data. Availability and provider terms remain separate from the core skill.

// COMMAND CONTRACT

What does the audit require and write?

Complete command surface

The audit skill has one documented user command and no audit-specific subcommands.

CommandRequired inputPurpose
/seo audit <url>A reachable website URLRun the full-site audit and create a prioritized local artifact package

Companion environment commands

CommandUse
/seo setupCreate or refresh Claude SEO's isolated runtime and Playwright Chromium when required.
/seo doctorCheck runtime, dependency, and Chromium readiness without changing the system.

Inputs, permissions, and cost boundaries

  • Network: the audit requests the supplied URL and reachable internal pages, subject to robots.txt, timeouts, redirects, and rate limits.
  • Local writes: it creates a {domain}-audit/ directory. The reviewed process does not include a step that changes the audited website.
  • Credentials: the core command can run without Google or DataForSEO credentials. Optional specialists activate only when their required credentials or services are available.
  • Costs: Claude SEO is MIT-licensed with no separate license fee. Claude Code access and optional providers can have their own usage costs, terms, and limits.

Representative workflow

This is an example sequence, not a captured audit result:

# Prepare or verify the runtime if needed
/seo setup
/seo doctor

# Run the full-site audit
/seo audit https://example.com

Documented outputs

ArtifactWhat it contains
{domain}-audit/FULL-AUDIT-REPORT.mdExecutive summary and category findings
{domain}-audit/ACTION-PLAN.mdCritical, high, medium, and low priority work
{domain}-audit/audit-data.jsonStructured audit envelope used for report generation
{domain}-audit/findings/*.mdPer-category specialist evidence and recommendations
{domain}-audit/screenshots/Desktop and mobile captures when Playwright is available
Optional PDF reportGenerated from the structured audit envelope when requested after completion
// SCORE + LIMITS

How should you interpret the result?

CONTENT QUALITY
23%
TECHNICAL SEO
22%
ON-PAGE SEO
20%
SCHEMA
10%
PERFORMANCE
10%
AI READINESS
10%
IMAGES
5%

The score is a weighted summary for triage. Read the underlying evidence and category findings before deciding what to change.

  • Blocked pages, unreachable URLs, rate limits, and the 500-page ceiling can make the result partial.
  • Core Web Vitals field status requires field data. Lab or heuristic findings do not prove a field pass.
  • Screenshots depend on a working Playwright route. Missing captures must be reported, not silently treated as passed.
  • Google, backlink, maps, and DataForSEO enrichment depends on available credentials and provider responses.
  • Business-type detection is automated. Verify the classification before accepting industry-specific recommendations.
  • An audit identifies and prioritizes work. It does not guarantee ranking improvement, traffic, conversion, or indexation.
// VERIFIED SOURCES

SOURCE-CHECKED

Tool behavior and commands were checked against a dated Claude SEO source snapshot. Search guidance and Claude Code access requirements were checked against current first-party documentation. Verified 2026-08-26.

The pinned skill file is a product contract, not proof that a live audit succeeded. This page does not claim a runtime, score, or result from an audit that was not run.

// FAQ

What should you know about the SEO audit?

The command does not promise a fixed runtime. Duration depends on how many pages are reachable, rendering work, rate limits, and available integrations. The crawl configuration allows up to 500 pages.
// RELATED SKILLS

EXPLORE MORE

VIEW ALL 25 SKILLS →

RUN YOUR FIRST AUDIT
WITH SPECIALIST REVIEW.

$
git clone --depth 1 https://github.com/AgriciDaniel/claude-seo.git && bash claude-seo/install.sh
VIEW SOURCE ALL SKILLS >