Skip to content
CiteScore

Guide

How to get cited by Perplexity Comet

What Perplexity publishes for Comet’s browser assistant: it reads the open page’s visible text, headlines, and metadata. Search citations still use PerplexityBot. No separate Comet token, and no citation rate.

21 October 2026 · CiteScore Team · 11 min read

Comet is Perplexity’s browser. The assistant sits in a panel beside the page a person is already reading. Perplexity’s assistant panel article calls that assistant a Perplexity service, and says it can pull the open page, or a tab the person names with @. That is a different path from Perplexity search, where Perplexity’s crawler documentation tells a publisher to allow PerplexityBot. CiteScore does not open Comet. The claims below are from those two pages, the Comet assistant privacy article, and Perplexity’s how-it-works page, read on 21 October 2026. None of them publishes a citation rate.

The search-crawler page, including the firewall rule and the user-triggered fetcher, is how to get cited by Perplexity. This page stays on the browser assistant. The same kind of split for other products is written up for ChatGPT, Claude, Gemini, Microsoft Copilot, Grok, Meta AI, DeepSeek, Mistral, Brave Search, You.com, and DuckDuckGo. The short map is how to get cited by AI answer engines.

The open page is what the assistant can read

The assistant panel article says the assistant runs beside the main browser window. It describes the assistant as context-aware: it can use the tab the person is viewing, or a tab they mention with @. The section on context extraction says the assistant pulls in the visible text, the headlines, and the website metadata from the open page every time the person submits a request. A summary, a question about that page, and an action on the site are the uses that article names.

“Visible text” and “headlines” are the words that page uses. A sentence that exists only inside a script, or only after a control the person has not opened, is not text that description covers. Put the answer in the HTML the browser shows: what the thing is, who provides it, and the fact a question would quote, in the title and the first paragraph, and again in a heading when the heading is the question. The article also says the assistant pulls website metadata. It does not list which tags. A title and a meta description are metadata a publisher can see in the page source. This guide does not turn “metadata” into a list of headers Perplexity did not publish.

The privacy article says Comet Assistant helps a person summarize pages, automate tasks, and retrieve information. By default it does not upload browsing history, the full list of open tabs, cookies, passwords, local files, or text the person typed into a site unless they send that text in a request. It sends page content when the request needs it, including a summary of the current tab, selected text, or a tab named with @tab. The article says only the context the request needs is sent. A publisher cannot flip that send on for someone else’s browser. The person asks.

Search citations still use PerplexityBot

The assistant is a Perplexity service. When the question is a search of the web, the crawler page is the control Perplexity publishes. PerplexityBot is the user agent that page says surfaces and links websites in Perplexity search results. Perplexity recommends allowing it in robots.txt and permitting the published IP ranges. The full group, the WAF note, and the up-to-24-hour delay are on how to get cited by Perplexity. The token table is the AI crawler user-agent reference. How a disallow is written is the robots.txt guide.

That crawler page also names Perplexity-User. It says this fetcher supports a user action: when someone asks Perplexity a question, it might visit a page and include a link. It says the fetcher generally ignores robots.txt because a person requested the fetch. The Comet articles read here do not say the assistant panel is that fetcher, and they do not say it is a different one. This guide does not invent the missing sentence. A page PerplexityBot is forbidden to crawl is a page the crawler documentation does not promise to surface in search. An open tab is the other path: the person already loaded it, and the assistant panel article describes reading what is visible.

Perplexity’s how-it-works page says each answer includes numbered citations linking to the original sources. That page is about Perplexity’s answers. It is not the assistant panel article, and it does not mention Comet. A numbered citation in a Perplexity answer and a summary of a tab the person already has open can both happen inside a Perplexity product. They are not the same claim, and this page does not publish a rate for either one.

What the Comet articles leave out

The assistant panel article and the privacy article do not list a schema type, an llms.txt file, a word count, or a freshness score. The privacy article says assistant context that includes page content can be stored for up to 30 days for Library and query history, and then deleted. That is retention. It is not a statement that a page updated 30 days ago is preferred, and this guide does not turn it into one. A visible updated date is still a fact a person can check on the page. These articles do not say Comet ranks by it.

The privacy article also says a person can turn the assistant off, or block it on specific websites, under Settings, Privacy and security, Comet Assistant. That control belongs to the person using Comet. A publisher does not get a switch in robots.txt from that paragraph. Blocking the assistant on your own browser does not remove your site from Perplexity search. Allowing PerplexityBot does not force someone else’s assistant to open your URL.

What to change on the page

  1. Keep the URL publicly fetchable if you want it eligible for Perplexity search. The crawler page recommends allowing PerplexityBot. A login wall is a page the search crawler cannot quote from the public HTML. The open-tab path is separate: the assistant reads the visible text of a page the person has open.
  2. Leave PerplexityBot able to fetch the path when you want the site in Perplexity search results. Do not add a Comet user-agent group. The pages read here do not name one. The search write-up is how to get cited by Perplexity.
  3. Write the answer in the visible text and in the headlines. The assistant panel article says those are what the assistant pulls from the open page, along with website metadata. Perplexity does not publish a word count, and this page does not invent one.
  4. Repeat the visible facts in JSON-LD if you want the labels. A recipe page can use the Recipe schema generator. The field list is the schema guide. The Comet articles do not require either script. The markup has to match the copy a person can see.
  5. Put a visible updated date on the page when the page has one. The Comet articles do not publish a freshness score. A date the page shows is still the date a reader can check. Do not invent a “recently updated” line to chase a recrawl interval these pages do not state.
  6. Add /llms.txt only if some other reader should have a curated list. The llms.txt generator and the llms.txt best practices page describe that file. The Comet articles read here do not ask a publisher for it.

Free AI Answer Snapshot, then the Fix Pack

The free AI Answer Snapshot asks logged-out ChatGPT and Perplexity who they name for a local business and sends one page. AI answers vary run to run. No payment. The Fix Pack, $49 is the done-for-you work after you have seen who was named.

The free check reports GPTBot, PerplexityBot, and Google-Extended. It does not open Comet, and it does not look for a Comet token. A clear PerplexityBot row is not a Comet citation. The GEO audit scores one public URL from the HTML we fetch and returns about ten fixes. It does not open Comet or Perplexity.

Run the free check on the host. It reads the homepage, robots.txt for GPTBot, PerplexityBot, and Google-Extended, /llms.txt, a sitemap, and JSON-LD on the homepage. It does not log into an answer engine. Paid files are listed at checkout.

Questions about Comet citations

Does Comet have its own robots.txt user agent?

Not on the pages read for this guide. Perplexity’s crawler documentation, read on 21 October 2026, names PerplexityBot and Perplexity-User. It does not name a Comet token. The Comet assistant panel article and the Comet assistant privacy article do not name one either. PerplexityBot is the token Perplexity says to allow when you want a site in Perplexity search results. A CometBot group would be a token this guide invented, so this page does not publish one.

Does Comet’s assistant cite the page a person already has open?

The assistant panel article says the assistant pulls the visible text, headlines, and website metadata from the open page each time a person submits a request, and that it can also use a tab the person names with @. That article does not say this extraction is published as a numbered citation to other people. Perplexity’s how-it-works page says Perplexity answers include numbered citations to sources. Those are two descriptions. This guide does not merge them into one rate, and it does not claim the open-tab summary is the same event as a search footnote.

Does Perplexity publish a citation rate, a required schema type, or a freshness score for Comet?

Not on the assistant panel article, the assistant privacy article, the crawler documentation, or the how-it-works page read for this guide. Those pages do not list a schema type, an llms.txt file, a word count, or a citation rate. JSON-LD that matches the visible page can still help other readers. It is not a switch on these pages. The privacy article’s 30-day figure is a retention window for assistant context, not a ranking interval.

Does the free check query Comet?

No. The free check reads GPTBot, PerplexityBot, and Google-Extended, plus the homepage, llms.txt, a sitemap, and JSON-LD on the homepage. It does not open Comet, and it does not look for a Comet user agent, because the pages read here do not publish one. A clear PerplexityBot row means that robots group is not a disallow. It is not a Comet citation.

Sources

Request a free AI Answer Snapshot

Same request as the free snapshot page. AI answers vary run to run. This is a snapshot, not a ranking, and not a promise that any assistant will name the business.

All fields are required.

We email the snapshot to this address. We do not ask for passwords, cookies, or content behind a login. We do not sell emails. Privacy.

Email this request instead
citescore@mail.grokbot.com

No mail app opened? Copy this and email it to citescore@mail.grokbot.com from any inbox.

If that send fails, email this request instead, or use the copy buttons.

Sends the business name, city, service, website, and email to citescore@mail.grokbot.com through FlowForm. This page path is included. No account and no paid checkout.

Prepared by the CiteScore Team.