How this score is measured
The overall score weights Live Answer Visibility at 45%, Search Index Foundation at 30% and Training Data Presence at 25%, rounded to the nearest point. Critical findings can cap the score lower; any applied cap is stated next to it. Live Answer Visibility and Training Data Presence reflect crawler access plus on-page structure, with Training Data Presence also counting Common Crawl. Search Index Foundation reflects Googlebot and bingbot access plus indexability, sitemap and response health.
Our crawler tests send each bot's user agent from our own servers, not from the AI companies' verified IP ranges. Some firewalls, including Cloudflare, verify crawlers by network identity rather than user agent. That means a site can challenge our test request while allowing the real, verified crawler. When that happens this scan over-reports blocking; it never under-reports it. The Googlebot result is the most exposed to this, because fake Googlebots are the most common spoof and are policed hardest.
This scan of replit.com checked 15 AI crawlers across 21 automated checks on 6 pages. None of the 13 scored crawlers are blocked. AI Access Score: 97/100 (Fully open).
The bot matrix
robots.txt policy next to what actually happened when each bot knocked. Rows where your policy says yes but the server says no are highlighted; those are the silent failures.
Live-answer botsFetch content in real time to build the answers and citations users see today.
| Bot | Owner | robots.txt policy | Real fetch | Verify |
|---|---|---|---|---|
| OAI-SearchBot Powers ChatGPT search results | OpenAI | Allowed | Reachable (200) | |
| ChatGPT-User User-triggered fetches from ChatGPT; Cloudflare classes this as Agent traffic, blocked by default on ad pages for new Cloudflare domains | OpenAI | Allowed | Reachable (200) | |
| PerplexityBot Index for Perplexity answers | Perplexity | Allowed | Reachable (200) | |
| Perplexity-Userinformational User-triggered; may ignore robots.txt | Perplexity | Allowed | Reachable (200) | |
| Claude-SearchBot Claude search indexing | Anthropic | Allowed | Reachable (200) | |
| Claude-User User-triggered fetches from Claude; Cloudflare classes this as Agent traffic, blocked by default on ad pages for new Cloudflare domains | Anthropic | Allowed | Reachable (200) | |
| DuckAssistBot DuckAssist answers | DuckDuckGo | Allowed | Reachable (200) |
Training crawlersCollect content for model training. Blocking them removes you from future models' memory.
| Bot | Owner | robots.txt policy | Real fetch | Verify |
|---|---|---|---|---|
| GPTBot Training corpus | OpenAI | Allowed | Reachable (200) | |
| ClaudeBot Training corpus | Anthropic | Allowed | Reachable (200) | |
| CCBot Feeds many labs' training data; blocking it is the quietest way to disappear from future models | Common Crawl | Allowed | Reachable (200) | |
| Meta-ExternalAgent Llama training | Meta | Allowed | Reachable (200) | |
| Amazonbot Alexa and Amazon AI features | Amazon | Allowed | Reachable (200) | |
| Bytespiderinformational Known to sometimes ignore robots.txt | ByteDance | Allowed | Reachable (200) |
Search-index foundationCopilot rides Bing's index and AI Overviews rides Google's. If these starve, the AI layer above them starves.
| Bot | Owner | robots.txt policy | Real fetch | Verify |
|---|---|---|---|---|
| Googlebot Google Search index; AI Overviews rides this. Cloudflare treats it as mixed Search and Training, so as of September 15, 2026 Cloudflare AI training blocks can hit it | Allowed | Reachable (200) | ||
| bingbot Bing index; Copilot and ChatGPT search lean on it. Cloudflare treats it as mixed Search and Training, so as of September 15, 2026 Cloudflare AI training blocks can hit it | Microsoft | Allowed | Reachable (200) |
Training opt-out tokensNot crawlers. Directives controlling whether already-crawled content trains AI.
| Token | Controls | Your robots.txt |
|---|---|---|
| Google-Extended | Gemini training use of Google's crawl Blocking Google-Extended does NOT affect your Google Search ranking, but it does remove your content from Gemini training use. | Allowed |
| Applebot-Extended | Apple AI training use Blocking Applebot-Extended does not affect Siri or Spotlight search. | Allowed |
Your scanned pages
Real defects live on money pages, not just homepages. We scanned 6 pages: your homepage plus pages picked from your sitemap, preferring product, pricing and services pages, because those are the pages AI assistants quote when buyers ask.
| Page | Status | Indexable | Canonical | Title | Words |
|---|---|---|---|---|---|
| / | 200 | Yes | OK | AI App and Website Builder | Replit | 1432 |
| /pricing | 200 | Yes | OK | Pricing and Plans | Replit | 445 |
| /languages | 404 | Yes | None | missing | 3 |
| /languages/clojure | 200 | Yes | OK | Clojure Online Compiler & Interpreter… | 200 |
| /languages/haskell | 200 | Yes | OK | Haskell Online Compiler & Interpreter… | 196 |
| /languages/kotlin | 200 | Yes | OK | Kotlin Online Compiler & Interpreter | Replit | 199 |
Warnings (2)
Dead URLs in the crawl path waste crawler budget and erode trust in the sitemap.
https://replit.com/languages -> 404
curl -sI "https://replit.com/languages" | head -1Without a machine-readable identity, AI systems have to guess who you are instead of knowing.
Good to know (3)
A canonical tag prevents URL variants from splitting the page's identity.
curl -s "https://replit.com/languages" | grep -ioE '<link[^>]*rel=.?canonical[^>]*>'llms.txt is an emerging convention some AI crawlers read for site guidance. Yours is in place and follows the expected structure.
curl -s "https://replit.com/llms.txt"Copilot and ChatGPT search lean on Bing's index; if Bing has not indexed you, two of six AI platforms cannot cite you. Automated Bing verification is unreliable without authenticated APIs, so we will not fake it.
Recommended robots.txt additions (free, no email needed)
# Recommended AI crawler access, generated by the AI Crawler Access Report by Citant.ai # Place these lines ABOVE any general User-agent: * block in robots.txt User-agent: OAI-SearchBot Allow: / User-agent: ChatGPT-User Allow: / User-agent: PerplexityBot Allow: / User-agent: Claude-SearchBot Allow: / User-agent: Claude-User Allow: / User-agent: DuckAssistBot Allow: / User-agent: GPTBot Allow: / User-agent: ClaudeBot Allow: / User-agent: CCBot Allow: / User-agent: Meta-ExternalAgent Allow: / User-agent: Amazonbot Allow: / User-agent: Googlebot Allow: / User-agent: bingbot Allow: / # Optional training opt-outs. These are business decisions, not defects: # blocking Google-Extended removes you from Gemini training but does NOT # affect Google Search ranking. Leave them out to stay fully open. # User-agent: Google-Extended # Disallow: /
Your prioritized fix plan
No criticals found. Warnings are ordered by impact. Platform detected: Next.js. Open the print-ready report to save it as a PDF.
1. Fix your sitemap
Fixes: Sitemap or key pages return errors (1 sampled)
- Ensure /sitemap.xml returns 200 and lists your live URLs; remove entries that 404.
- Reference it from robots.txt with a Sitemap: line so crawlers find it without guessing.
Sitemap: https://yourdomain.com/sitemap.xml
curl -sI "https://replit.com/languages" | head -12. Add Organization schema with sameAs links
Fixes: No Organization schema on the homepage
- Add the JSON-LD below to your homepage head, filling in your real details.
- The sameAs links are the entity-corroboration signal that connects your site to your LinkedIn and other profiles.
- Validate at validator.schema.org.
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Organization",
"name": "Replit",
"url": "https://replit.com",
"logo": "https://replit.com/logo.png",
"description": "What Replit does, in one sentence.",
"sameAs": [
"https://www.linkedin.com/company/replit",
"https://x.com/replit"
]
}
</script>3. Fix the canonical tag (/languages)
Fixes: No canonical tag on /languages
- The canonical must be an absolute https URL on this same host. A canonical pointing at a staging domain or another host donates the page's identity elsewhere.
- Fix the template that renders the link rel=canonical tag, deploy, and verify with the command in this report.
curl -s "https://replit.com/languages" | grep -ioE '<link[^>]*rel=.?canonical[^>]*>'