What Is PerplexityBot, and How Do You Let It Past a Firewall?

PerplexityBot vs Perplexity-User

PerplexityBot and Perplexity-User are the two agents named in Perplexity’s crawler documentation. The first crawls ahead of time so your pages can appear and be linked in Perplexity’s results. The second visits a page only when someone’s question needs it, then may link that page in the answer.

The difference shapes what a robots.txt block can do. A block on the indexing bot is respected. A Perplexity-User visit, triggered by a person, usually is not.

OpenAI draws a similar line with ChatGPT-User, saying robots.txt rules “may not apply” to user-started visits. Our AI crawler entry compares these fetchers across vendors.

PerplexityBot user agent strings and published IP ranges

PerplexityBot announces itself with a short, fixed user-agent string, and Perplexity publishes a separate IP list for each of its two agents. Use both pieces together. A string tells you who claims to be visiting, while the IP list tells you whether the claim is true.

Perplexity’s crawler page gives these full strings:

Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; PerplexityBot/1.0; +https://perplexity.ai/perplexitybot)

Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; Perplexity-User/1.0; +https://perplexity.ai/perplexity-user)

The IP lists sit at perplexity.com/perplexitybot.json and perplexity.com/perplexity-user.json.

On 26 October 2026 the PerplexityBot file held 8 IPv4 addresses and the Perplexity-User file held 4. Perplexity says the addresses “are updated regularly”.

So treat the files as live data. Perplexity itself suggests automated jobs that fetch them on a schedule and refresh your rules.

A quick log routine shows both halves at once. Pull every request carrying either token, then count the source addresses:

grep -E "PerplexityBot|Perplexity-User" access.log | awk '{print $1}' | sort | uniq -c | sort -rn

Each address should sit in the matching JSON file. Digital Magma flags any that do not, because a borrowed name from an unknown network is a scraper, whoever it claims to be.

Keep the two tokens apart while you read the output. Index crawls and live user fetches have different rules, so mixing them hides problems.

PerplexityBot robots.txt rules and timing

PerplexityBot rules work like any other named robots.txt group. Allow it to stay in Perplexity’s results, or disallow it to leave. Perplexity’s crawler page currently gives no figure for how quickly a change takes effect, so check your logs instead of assuming a fixed delay.

# Stay visible in Perplexity
User-agent: PerplexityBot
Allow: /
Disallow: /wp-admin/

# Leave Perplexity's index
User-agent: PerplexityBot
Disallow: /

Remember the * group trap. Under RFC 9309, a named group replaces the * group for that crawler, so private paths must be repeated.

A Perplexity-User group is legal syntax, yet Perplexity says that fetcher generally ignores robots.txt. Use authentication for pages that must stay private.

On timing, RFC 9309 says crawlers should not use a cached robots.txt for more than 24 hours. That is the standard, not a Perplexity promise.

Digital Magma rechecks logs a few days after any change. That shows whether crawler requests stopped or resumed as intended.

Perplexity through Cloudflare and AWS WAF: allow rules

Perplexity’s documentation includes firewall recipes, which is unusual and helpful. For Cloudflare and AWS WAF, it asks owners to build allow rules that combine a user-agent match with an IP-range match. A rule built on the name alone would also wave through anyone spoofing it.

Cloudflare. Perplexity’s steps, in short:

1. Go to Security, then WAF, then Custom rules.
2. Field User Agent, operator Contains, value PerplexityBot or Perplexity-User.
3. Add an AND condition: IP Source Address is in the official ranges.
4. Set the action to Allow.

AWS WAF. Create IP sets for both agents and string-match conditions for both user agents. Combine them in allow rules, attach them to your Web ACL, and give them higher priority than blocking rules.

Perplexity adds that WAF changes “may take some time to propagate”. Watch the logs to confirm real Perplexity traffic gets through.

PerplexityBot access for a business: why it matters

PerplexityBot access decides whether Perplexity can link your pages in its results, and Perplexity answers usually show their sources. For a business, that makes the crawler a referral channel. Blocking it trades a source of qualified visits for a training concern that Perplexity says does not apply.

That last point deserves weight. Perplexity states PerplexityBot “is not used to crawl content for AI foundation models”.

So a business that blocks the bot to stop AI training gains nothing on that front. It only removes itself from Perplexity’s index.

Digital Magma allows PerplexityBot and Perplexity-User on its own site, each in a named group. Our Perplexity entry explains how the answer engine chooses and shows sources.

Firms that do have licensing concerns should still separate the decisions. Set training rules per vendor, then judge Perplexity on search value alone.

There is a measurement angle too. Perplexity links its sources, so visits it sends usually arrive as referrals that analytics can count.

Digital Magma sets those referrals up as their own channel in GA4 for clients. Without that, the visits blend into general referral traffic and nobody can judge the value of staying open.

PerplexityBot in our client work

PerplexityBot access is part of every AI visibility review Digital Magma runs for clients. For a Canadian welding equipment manufacturer already visible in ChatGPT and Google’s AI Overviews, the same open access lets answer engines like Perplexity reach its detailed product pages.

Engineers researching equipment tend to ask long, specific questions. Answer engines that show sources reward pages with exact specifications, which that manufacturer’s site has.

For a Toronto ENT clinic, the firewall matters more. Clinic sites often sit behind strict security presets that challenge unknown bots.

For an app development company serving six countries, the check repeats on every regional host and CDN zone.

One forgotten zone with an aggressive bot preset is enough to hide a whole market from the crawler.

Perplexity crawler disputes: what site owners can check for themselves

Perplexity’s crawling became a public dispute in August 2025. Cloudflare published a report saying Perplexity reached blocked content through undeclared crawlers. Perplexity responded publicly on its own blog. Site owners do not need to take either side on trust, because their own logs can answer the question.

Cloudflare’s report, dated 4 August 2025, describes tests on new domains blocked by robots.txt and WAF rules.

It says Perplexity still answered questions about them, and that it saw a generic Chrome-on-macOS user agent from IPs outside Perplexity’s published ranges.

Perplexity’s reply post blocked our automated fetch, so we do not quote it here. Read it on Perplexity’s blog alongside Cloudflare’s report.

What you can check yourself:

• Do declared PerplexityBot requests come only from listed IPs?
• Do requests stop on paths you have disallowed?
• Does your CDN’s bot report flag unverified traffic on blocked pages?

PerplexityBot: frequently asked questions

PerplexityBot questions from site owners and developers, answered from Perplexity’s documentation.

PerplexityBot: related terms, sources and author

PerplexityBot ties into these Digital Magma entries:

• Perplexity: how the answer engine picks sources
• AI crawler: training, search and user fetchers
• Robots.txt for AI bots: copy-ready rule sets
• ClaudeBot: Anthropic’s three bots
• How to get cited by ChatGPT and Perplexity and whether to block AI crawlers

Sources: Perplexity’s “Perplexity Crawlers” page and IP lists, Cloudflare’s August 2025 report, and IETF RFC 9309. Checked on 26 October 2026.