AI Crawler Audit — Are You Blocking ChatGPT?

Check whether your site is visible to AI search — or quietly blocking the crawlers that feed ChatGPT, Perplexity, Claude and Google's AI Overviews. Audits robots.txt for 12 AI crawlers, detects llms.txt, and returns an AI-readiness score with the exact fixes.

No API key, no login, no anti-bot — these are public files every site already serves.

The problem this catches

The old SEO reflex is block the scrapers. The new GEO reality is be citable. So sites routinely block GPTBot and PerplexityBot in robots.txt while paying an agency to improve their AI visibility. This actor finds that contradiction in seconds, across a whole client list.

What you get

Field	Description
domain	The site audited
aiReadinessScore	0–100: can AI assistants actually read and cite you?
aiSearchReadiness	optimized · open · partial · blocked
blocksAiSearch	true if any answer-critical crawler is blocked
blockedSearchBots	Which answer-critical bots are blocked (the ones that cost you citations)
blockedBots	Every blocked AI crawler, including training-only ones
hasLlmsTxt / llmsTitle / llmsSections / llmsLinks	Your llms.txt, if you publish one
allowedBotCount / totalBotsChecked	Coverage of the 12 crawlers checked
sitemapCount / robotsStatus	robots.txt health
recommendations	The specific fixes, in plain language
botDetail (optional)	Per-bot verdict + the exact robots.txt rule that matched

Crawlers checked: GPTBot · OAI-SearchBot · ChatGPT-User · ClaudeBot · Claude-User · PerplexityBot · Perplexity-User · Google-Extended · Applebot-Extended · CCBot · Bytespider · meta-externalagent.

How to use it

1. Paste your domains (or a whole client list — full URLs are fine).
2. Run it. Sites are returned worst score first, so the problems are at the top.
3. Tick Only return sites that block AI search when auditing a long list — you get just the ones with a real problem.

Input

{
  "domains": ["stripe.com", "notion.so", "example.com"],
  "onlyBlockedSites": false,
  "includeBotDetail": false,
  "maxItems": 1000
}

Sample output

[
  {
    "domain": "example.com",
    "aiReadinessScore": 14,
    "aiSearchReadiness": "partial",
    "blocksAiSearch": true,
    "blockedSearchBots": "OAI-SearchBot, PerplexityBot, Google-Extended",
    "blockedBots": "GPTBot, OAI-SearchBot, PerplexityBot, Google-Extended, CCBot",
    "hasLlmsTxt": false,
    "llmsTitle": null,
    "allowedBotCount": 7,
    "totalBotsChecked": 12,
    "sitemapCount": 1,
    "robotsStatus": 200,
    "recommendations": "Unblock OAI-SearchBot, PerplexityBot, Google-Extended in robots.txt — these decide whether you can appear in AI answers. | Add an /llms.txt so AI assistants get a curated map of your key pages.",
    "status": "ok"
  }
]

Typical uses

• Agency audit — run a client list and lead with "you're invisible to ChatGPT, here's the one-line fix."
• Pre-GEO baseline — check AI-crawler access before paying for AI-visibility work; pairs directly with the AI Brand Visibility Tracker actor (this says can they read you, that says do they cite you).
• Competitor check — see which competitors have opened up to AI crawlers and published llms.txt.
• Monitoring — schedule it; a robots.txt edit that silently blocks GPTBot is a real regression.

Pricing

$1.00 / 1,000 sites audited ($0.001 per result), plus a near-zero start fee. Never charged beyond your maxItems; unreachable domains are reported as rows with a failureReason, not billed as data.

FAQ & limitations

• Does a missing robots.txt mean blocked? No — the opposite. Crawlers fall back to allow-all, which is unmanaged rather than blocked; the actor flags the non-200 status so you can decide.
• Is llms.txt required? No, it's an emerging convention. Publishing one adds 15 points here because it gives assistants a curated map of your key pages.
• Does blocking a training bot hurt citations? Not directly — that's why the score weights the answer-critical bots (search + live-fetch) and lists training-only blocks separately.
• Does this prove you'll be cited? No. It proves you're reachable. Use the AI Brand Visibility Tracker to measure whether you're actually mentioned.
• Rate limits / blocks? robots.txt is public and cheap to fetch; the proxy toggle exists but is rarely needed.

Related actors

• AI Brand Visibility Tracker — do ChatGPT/Perplexity actually mention and cite you?
• HTTP Header Analyzer — security-header grade for the same domains.
• Broken Link Checker · Sitemap URL Extractor — the rest of a technical site audit.