Skip to content
RFC 9309 · longest-match tester · AI-crawler registry · live lint

Robots.txt Studio

Build a robots.txt with AI-crawler policy presets and lint it as you type — then test any URL against any file with real RFC 9309 longest-match semantics, with the winning line highlighted. 100% in-browser, no AI, no uploads.

1 group1 rule0 sitemap lines0.0 KiB / 500 KiB cap
Live lint
No syntax issues found.
Start from a preset
Add an AI-crawler policy
Declare your sitemap
robots.txt is a request, not enforcement. Bytespider frequently ignores it, and user-triggered AI fetchers (ChatGPT-User, Claude-User, Perplexity-User) act on behalf of humans and may bypass it by design. Hard blocks need server or WAF rules.
Registry

AI crawler user-agents — one-click policy

OperatorUser-agentPurposerobots.txtIn your file
OpenAIGPTBottraininghonors
OpenAIOAI-SearchBotAI searchhonors
OpenAIChatGPT-Useruser-triggeredhonors
AnthropicClaudeBottraininghonors
AnthropicClaude-SearchBotAI searchhonors
AnthropicClaude-Useruser-triggeredhonors
PerplexityPerplexityBotAI searchdeclared
PerplexityPerplexity-Useruser-triggered fetchers may act on behalf of humans and can ignore robots.txt by designuser-triggeredmixed history
GoogleGoogle-ExtendedGemini training opt-out token — does NOT affect Google Searchtraininghonors
GoogleGoogle-CloudVertexBottraininghonors
AppleApplebot-Extendedtraining opt-out token; Applebot itself powers Siri/Spotlight searchtraininghonors
Common CrawlCCBotopen dataset that feeds many LLMstrainingmixed history
MetaMeta-ExternalAgenttrainingdeclared
MetaMeta-ExternalFetcheruser-triggereddeclared
ByteDanceBytespiderfrequently ignores robots.txt — server/WAF rules needed for a hard blocktrainingoften ignores
AmazonAmazonbottraininghonors
DuckDuckGoDuckAssistBotAI searchhonors
MistralMistralAI-Useruser-triggereddeclared
(legacy)anthropic-aideprecated token, kept for compatibility blockstraininghonors
(legacy)Claude-Webdeprecated token, kept for compatibility blockstraininghonors

Registry v1.0 (reviewed quarterly). Buttons edit standalone User-agent blocks in the Build editor above.

Reference

RFC 9309 matching rules

Longest match wins

Among all Allow/Disallow rules that match a path, the one with the most matched characters decides — file order is irrelevant.

Allow wins ties

If an Allow and a Disallow match with equal length, Google resolves the tie in favor of Allow.

* and $ wildcards

* matches any character sequence; $ anchors the end of the URL. Disallow: /*.pdf$ blocks PDFs but not /file.pdfx.

Most-specific group

A crawler obeys only the group whose User-agent token best matches its name; * applies only when nothing else does.

Field notes

Robots.txt in 2026: crawl control meets AI policy

robots.txt is the oldest contract on the web — a plain-text file at the site root that asks crawlers what not to fetch — and in 2022 it finally became a real standard, RFC 9309. The formalization matters because the folk understanding (“rules apply top to bottom”) is wrong: compliant crawlers pick the most specific user-agent group, then apply the longest matching rule, with Allow winning ties. That's why a one-character change to a path pattern can flip a verdict, and why this studio tests with the actual algorithm and shows you the exact winning line rather than asking you to reason about it.

The stakes changed with AI crawlers. Beyond Googlebot and Bingbot, your file now speaks to training bots (GPTBot, ClaudeBot, CCBot, Meta-ExternalAgent), AI-search indexers (OAI-SearchBot, Claude-SearchBot, PerplexityBot) and user-triggered fetchers (ChatGPT-User, Claude-User). These are genuinely different policy decisions: many sites want to be findable in AI search while opting out of model training — which is exactly what the “search-visible, no training” preset emits, built from the bundled registry. Google-Extended deserves special mention: blocking it opts you out of Gemini training without touching your Google Search presence.

Be honest about what the file can and cannot do. It is a request, not a lock: Bytespider has a documented history of ignoring it, and user-triggered fetchers act on behalf of a human who could open your page in a browser anyway. If you need enforcement, that's a server-level or WAF decision. Also remember two classic traps: a page blocked in robots.txt can still appear in results (Google just can't read it — including any noindex tag on it), and noindex inside robots.txt has been ignored by Google since 2019.

Keep the file lean — Google parses only the first 500 KiB — declare your sitemap with a Sitemap: line so discovery doesn't depend on links alone, and re-test after every edit. Then verify the rest of the stack: run the full SEO Audit Pro on key pages, build the sitemap itself in the XML Sitemap Studio, and pair your AI policy with an llms.txt so the crawlers you allow know where the good content is.

Robots.txt Studio FAQs

Have more questions? Contact us

Trusted for Crawl Control

4.9
Based on 2,740 reviews

The winning-line highlight is the feature every robots tester should have had years ago. I batch-tested 80 URLs from a migration against the new file and caught two patterns where a shorter Disallow was being beaten by an old Allow. The RFC semantics are actually correct, which I cannot say for most testers.

S
Sofia Andersson
Technical SEO lead
June 14, 2026

We used the 'search-visible, no training' preset as our AI policy baseline across 14 properties. Having the registry with honest honors/ignores labels — including the Bytespider caveat — saved me a day of research. The lint caught a rule sitting above our first User-agent line that every crawler had been ignoring.

D
Dev Kulkarni
Platform engineer
May 30, 2026

Client staging site got indexed because of a robots typo — now I lint every file here before deploy. The byte counter against Google's 500 KiB cap and the BOM detection are details nobody else bothers with. Paste, test, ship. No signup, nothing uploaded.

R
Rachel Odei
SEO consultant
April 22, 2026

The e-commerce facet preset was 90% of what our store needed; I added two site-specific patterns and verified them against real product URLs in test mode. Would love saved profiles someday, but the export is clean and the tester's reasoning strings make code review of robots changes actually possible.

J
Jonas Weber
E-commerce developer
March 18, 2026

Love using our calculator?

Connected instruments

More of SEO Forge

Learn More

Related Articles

Dive deeper with our expert guides and tutorials related to Robots.txt Studio

Loading articles...

RFC 9309 longest-match tester · live lint · AI-crawler registry & presets · sitemap lines · in-browser · Last reviewed: 2026-07