Site & GEO Audit

Find what's broken. Your agent fixes it. LogNorm verifies.

A crawl-based SEO audit and a GEO audit for AI crawlers and agents, with every finding tied to its URL. Your agent fixes the cause in your codebase, and LogNorm re-checks the live pages until the move closes.

Works withClaude CodeClaude CodeCodexCodexCursorCursorClaudeClaudeand any MCP client
Site · acme.com · audit
78
SEO health
64
GEO score
412 pages crawled · 96% coverage Export all findings
  • criticalBroken internal links14 pages
  • criticalAnswer-engine crawlers blockedGEO1 site
  • warningMissing meta description23 pages
  • warningThin contentAI-judged84%6 pages
  • warningPages aren't available as markdown to AI agentsGEO1 site

Illustrative example. acme.com is a sample site.

Live in your workspace
  • Sam's Panda commented on Fix broken internal links
    Redirects added in next.config.ts
  • Sam's Panda picked up Fix broken internal links
    Claimed from this week's plan
  • LogNorm found 934 pages returning 404
    SEO audit · critical
Why it matters

Audits are easy. Fixing is the hard part.

01

Audits list problems nobody has time to fix.

Each issue becomes a move with its pages attached, ranked against the rest of your work.
02

Fixes live in the codebase, far from the SEO tool.

Your agent works in your repo: it finds where the pages come from and fixes the cause, not each page by hand.
03

Nobody checks whether the fix worked.

LogNorm re-fetches the affected pages on the live site and closes the move only when they pass.
Humans and agents, in parallel

Found, fixed and verified in one loop

You and your team
Decide, review, publish
  1. 01Decide which fixes to take this week
  2. 02Review the pull request
  3. 03Merge and deploy
Your agents
Claude Code, Codex, Cursor, Claude
Claude CodeCodexCursorClaude
  1. 01Read every affected URL and the rule
  2. 02Fix the template, metadata or redirects
  3. 03Ship it, then ask LogNorm to re-check
LogNorm
Finds, ranks, verifies, measures
  1. 01Crawl and audit on your schedule
  2. 02Check what AI crawlers and agents can read
  3. 03Re-check fixes on the live site
SEO audit

Every finding keeps its URL, so you can check it yourself

A rule only reports what it checked. noindex is confirmed page by page, with the meta tag and the X-Robots-Tag header kept apart. Link-graph rules are skipped when the crawl covered too little of the site, and the audit says so instead of guessing.
  • Export all affected URLs for any rule as CSV.
  • AI-judged checks are labelled as LogNorm's call, with a confidence.
  • Each issue becomes a move with its pages attached.
Findings · Broken internal links
critical14 pages link to missing pages
CSV
  • acme.com/blog/invoice-templates
    links to/templates/retainer404
  • acme.com/pricing
    links to/features/multi-currency404
  • acme.com/docs/getting-started
    links to/docs/api-keys410
  • + 11 more URLs in the export
Findings · Thin content
warningacme.com/integrations/zapier
180 words.Doesn't deliver its title84%
Skipped this audit: orphan pages and click depth. Crawl coverage was 84%, and link-graph rules need at least 90%.
Page types and the page-gap map

Know what every page is for, and which pages you're missing

Every crawled page gets a page type (product, solution, comparison, pricing, blog and more) and a funnel stage. The page-gap map lines up your pages against your competitors' along the buyer journey, and each empty stage is one click from a move.
  • Problem, solution, use case, integration, comparison, alternatives and pricing.
  • Competitor coverage comes from analysing their sites.
  • Each call shows its confidence and what it was based on.
Site · page-gap map
Buyer journeyYouGlobexInitech
ProblemAwareness3812064
SolutionConsideration496
Use caseConsideration073
IntegrationConsideration2145
ComparisonDecision062
AlternativesDecision031
PricingDecision111
acme.com/solutions/agencies
Solution93%Consideration81%
GEO audit

Check whether AI crawlers can reach you and quote you

The GEO audit reads robots.txt for each major AI crawler, tests whether your CDN refuses them, looks for llms.txt, and scores answerability and entity clarity across every indexable content page, up to 500 per audit. The GEO audit guide lists every check.
  • Search, user-triggered and training crawlers are reported separately.
  • Structured data, trust signals and freshness are checked too.
  • The lowest-scoring pages get suggested rewrites.
GEO audit · AI crawler access
  • OAI-SearchBot
    OpenAI · Search
    Allowed
  • Claude-SearchBot
    Anthropic · Search
    Allowed
  • PerplexityBot
    Perplexity · Search
    Blocked
  • ChatGPT-User
    OpenAI · User
    Allowed
  • GPTBot
    OpenAI · Training
    Blocked
/llms.txtNot found
AI answerability · every indexable page
  • acme.com/8/10
  • acme.com/pricing7/10
  • acme.com/features/recurring4/10
  • acme.com/blog/late-payments2/10
AI-judged. The lowest-scoring pages get suggested rewrites.
Agent readiness

See what AI agents find when they visit your site

Agents read and act through open standards. LogNorm checks markdown content negotiation, Content-Signal in robots.txt, RFC 8288 Link headers, the RFC 9727 API catalog and the protocol files agents look for: MCP server card, A2A agent card, Agent Skills index and AI catalog.
GEO audit · agent readiness
  • Markdown for agents
    Accept: text/markdown returns HTML
  • Content Signals
    search=yes, ai-input=yes, ai-train=no
  • Link headers
    No rel="describedby" or api-catalog link
  • API catalog
    /.well-known/api-catalog
  • MCP server card
    /.well-known/mcp/server-card.json
  • A2A agent card
    /.well-known/agent-card.json
  • OAuth discovery
    Shown for context, not scored
Build the fixes

Generate the files that fix what the audit found

After a GEO audit, a Build the fixes strip ranks the assets that would close the most findings. Generate the one you need, check it, and paste it into your site.
  • llms.txt, llms-full.txt and a robots.txt AI section.
  • A validated JSON-LD bundle, a fact sheet for AI assistants and the agent discovery kit.
  • Definition, FAQ, comparison and glossary pages as starting drafts.
GEO audit · Build the fixes
robots.txt AI sectionfixes 4 findings
Explicit rules per AI crawler, plus Content-Signal
Agent discovery kitfixes 3 findings
API catalog, Link headers, markdown-for-agents setup
llms.txtfixes 1 findings
A markdown map of your key pages
JSON-LD bundlefixes 3 findings
Organization, WebSite, SoftwareApplication, FAQPage
User-agent: *
Content-Signal: search=yes, ai-input=yes, ai-train=no

User-agent: PerplexityBot
Allow: /
Validate and schedule

Validate a fix without a full re-audit

Re-check one rule on the live site. Fixed findings are resolved, the rest keep their current message, and the move updates to what is left or closes when everything passes. Set a schedule, and the crawl and SEO audit run on their own.
  • Up to 1,500 affected pages re-fetched per validation.
  • Daily, weekly or monthly schedules per website.
  • Audit pages, exports and moves show only open findings.
Validate fix · Missing meta description
Re-checked 23 pages on the live site
20
Fixed
2
Still open
1
Not checked
The move now lists the 2 pages left. No full re-audit, no re-crawl of the other 389 pages.
Site · schedule
Crawl and SEO audit
Next run in 6 days
OffDailyWeeklyMonthly
Also included

An audit built to be acted on

29 SEO rules

Technical, metadata, content, indexability, links and schema, each with a severity.

Index status confirmed

For header-only noindex cases, LogNorm checks whether Google has the page indexed.

Per-page detail

Open any page to see its type, funnel stage, metadata, links and findings.

Score history

SEO health and GEO score over time, so you can see fixes land.

Findings become moves

Issues and page gaps join your Growth Plan, ranked against everything else.

Crawl coverage shown

Every audit states how much of the site it covered and what it skipped.
How it works

From crawl to fixed, in four steps

Details in the site audit docs and the GEO audit docs.

  1. 1

    Crawl

    Add your website. LogNorm crawls it, classifies every page and runs the SEO audit.
  2. 2

    Run the GEO audit

    Check AI crawler access, answerability, structured data and agent readiness.
  3. 3

    Fix

    Work the moves, export URLs for your team, or build the GEO assets.
  4. 4

    Validate

    Re-check one rule on the live site, or let the scheduled audit catch it.

SEO and GEO audit questions, answered

What does the SEO audit check?

LogNorm crawls your site and runs 29 rules: status codes and redirects, broken and redirected internal links, titles, meta descriptions and H1s, indexability and canonicals, thin and duplicate content, orphan and deep pages, image alt text, structured data on key pages, slow pages, sitemap, robots.txt and possible keyword cannibalisation.

How is a GEO audit different from an SEO audit?

An SEO audit asks whether search engines can crawl, index and rank your pages. A GEO audit asks whether AI assistants and agents can reach your pages, understand who you are and quote a clear answer: AI crawler access in robots.txt and at your CDN, llms.txt, structured data, entity clarity, answer-first content, trust signals and agent readiness.

Which findings are judged by AI, and how can I tell?

Thin content, keyword cannibalisation, page answerability and entity clarity need judgement rather than a count. Those findings are labelled as LogNorm's call and show its confidence. Everything else is a direct check of what the crawler fetched, such as a status code, a missing tag or a robots.txt rule.

Can I export the findings?

Yes. Every SEO rule and every GEO check exports all affected URLs as CSV, one rule at a time or everything at once. Exports only include findings that are still open.

Do I have to re-run the whole audit after fixing something?

No. Validate fix re-checks one rule on the live site: it re-fetches the affected pages (and link targets or duplicates where the rule needs them), marks fixed findings as resolved and updates the move with what is left. Robots and sitemap rules re-check the file. Orphan and depth rules need a full crawl, and site-wide or AI-judged GEO checks re-run the GEO audit.

What is agent readiness?

It checks what AI agents find when they visit your site through open standards: whether pages are served as markdown on request, whether robots.txt declares Content Signals, whether your home page sends useful Link headers, and whether you publish an API catalog, an MCP server card, an A2A agent card, an Agent Skills index or an AI catalog. OAuth discovery, auth.md, Web Bot Auth and DNS agent discovery are shown for context and not scored.

Can audits run on a schedule?

Yes. Per website, you can re-run the crawl and SEO audit daily, weekly or monthly on a paid plan or during the trial. The next run is counted from the last one, so a manual run pushes it out. If a scheduled run can't start, for example because you are out of credits, the reason is shown.

Can an AI agent fix the issues the audit finds?

Yes. Connect Claude Code, Codex or Cursor in the codebase that builds your site. It reads each finding with every affected URL, fixes the cause (templates, metadata, redirects, robots.txt, sitemaps, llms.txt), comments what it changed, and asks LogNorm to re-check the live pages. The move closes when everything passes.

Works with the rest of LogNorm

Audit your site for search and AI.

Start with a free growth scan, or create a workspace and run the full crawl, SEO audit and GEO audit.

Already use an agent?

Paste this into Claude Code, Codex, Cursor or Claude. It connects itself; you click Allow once.

Claude CodeClaude CodeCodexCodexCursorCursorClaudeClaudeand any MCP client