Short answer
Google-Agent is the user agent Google uses for AI agents hosted on its infrastructure that browse the web and take actions when a user asks. Google lists it as a user-triggered fetcher, so it generally ignores robots.txt. You can verify it against Google's user-triggered-agents.json IP list, and Google is testing Web Bot Auth signatures for it under the identity https://agent.bot.goog.
Key takeaways
- Google-Agent acts for one Google user, not for Google Search.
- Google says user-triggered fetchers generally ignore robots.txt.
- Verify it by IP list; treat signatures as an experimental extra.
What is Google-Agent?
Google-Agent is the name Google gives to AI agents running on its own infrastructure when they visit websites. Google's documentation says Google-Agent "is used by agents hosted on Google infrastructure to navigate the web and perform actions upon user request."
So when you see Google-Agent in your logs, a person has asked a Google-hosted agent to do something, and the agent came to your site to do it. That might mean reading a page, comparing options or taking an action.
Google lists Google-Agent among its user-triggered fetchers. These are fetchers that a user starts, as opposed to crawlers that run on their own schedule. Google's page does not name the specific Google products that use Google-Agent, so we don't guess here.
What does Google-Agent look like in my server logs?
It shows up as a browser-style user agent string that contains the token Google-Agent. Google publishes two versions, one mobile and one desktop.
Mobile:
Mozilla/5.0 (Linux; Android 6.0.1; Nexus 5X Build/MMB29P) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/W.X.Y.Z Mobile Safari/537.36 (compatible; Google-Agent; +https://developers.google.com/crawling/docs/crawlers-fetchers/google-agent)
Desktop:
Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko; compatible; Google-Agent; +https://developers.google.com/crawling/docs/crawlers-fetchers/google-agent) Chrome/W.X.Y.Z Safari/537.36
W.X.Y.Z stands for a Chrome version number, which changes over time. Don't match the whole string. Match the Google-Agent token instead.
Apart from the token, the string looks a lot like an ordinary Chrome browser. If you are looking for Google-Agent visits, search your raw server or CDN logs for the token.
Google adds a caution to its fetcher list: "The user agent string can be spoofed." Seeing the token tells you what a request claims to be, not what it is.
Does Google-Agent follow robots.txt?
Generally, no. Google says of all its user-triggered fetchers: "Because the fetch was requested by a user, these fetchers generally ignore robots.txt rules."
Google's page does not make a separate statement for Google-Agent, so this general rule is the best guidance available. A Disallow line aimed at Google-Agent is unlikely to stop it.
Robots.txt still matters for Google's crawlers, which work differently. For that side, see robots.txt for AI agents.
If you need a hard block, use your firewall or CDN rules, matched to Google's published IP list below.
How do I check a request is really from Google-Agent?
Check the IP address against Google's published list. Google says Google-Agent "uses IP ranges from user-triggered-agents.json", which lives at:
https://developers.google.com/static/crawling/ipranges/user-triggered-agents.json
If a request carries the Google-Agent token but comes from an IP outside that list, treat it as a request that claims to be Google-Agent but fails verification. Our article on spoofed AI crawlers covers how this works across vendors.
Two details are worth knowing:
- Reverse DNS. Google's verification page groups the user-triggered agents list with its user-triggered fetchers, whose hostnames match
***-***-***-***.gae.googleusercontent.comorgoogle-proxy-***-***-***-***.google.com. The IP list is the more direct check for Google-Agent. - The list changes. The file carries its own creation time and is updated over time. Fetch it on a schedule rather than copying the ranges once.
Does Google-Agent sign its requests?
Some of them, as a test. Google says it is "experimenting with the Web Bot Auth protocol, using the https://agent.bot.goog identity." Its Web Bot Auth page says "a subset of requests made by the Google-Agent are signed".
Signed requests carry Signature, Signature-Input and Signature-Agent headers, following HTTP Message Signatures (RFC 9421). The Signature-Agent header looks like this, with a g label:
Signature-Agent: g="https://agent.bot.goog"
The public keys are published at https://agent.bot.goog/.well-known/http-message-signatures-directory.
Google calls its implementation "experimental", partly because Web Bot Auth is still an IETF draft and partly because IP and user-agent checks remain the standard it is moving from carefully. It also warns: "Don't forget to fall back to IP-based verification, as not all requests are signed." So a missing signature is not a red flag on its own. A signature that is present but fails to verify is. For background on how these signatures work, see Web Bot Auth explained.
How is Google-Agent different from Googlebot and Google-Extended?
They do different jobs, and Google documents them separately.
| Name | What Google says it does | robots.txt |
|---|---|---|
| Googlebot | A common crawler used to build Google's search indexes. | Google says common crawlers always obey robots.txt when crawling automatically. |
| Google-Extended | A robots.txt control token for whether crawled content may be used for training Gemini models and for grounding. It has no separate user agent string. | It is itself a robots.txt rule. Google says it does not affect inclusion or ranking in Google Search. |
| Google-Agent | Used by agents hosted on Google infrastructure to navigate the web and take actions on user request. | A user-triggered fetcher; Google says these generally ignore robots.txt. |
The practical upshot: your Googlebot and Google-Extended rules don't decide what Google-Agent can do. Blocking Google-Extended in robots.txt is a choice about Gemini training and grounding. It is not a block on Google's agents visiting for a user.
Should I allow or block Google-Agent?
For most stores and business sites, allow it. Google-Agent visits because a real person asked an agent to do something on your site. That person may be comparing products, checking prices or getting ready to buy.
Reasons to be more careful:
- Sensitive areas. If parts of your site should never be used by an automated agent, such as account settings, protect them with the same logins and controls you use for any visitor. Robots.txt won't do it.
- Forms and actions. Google's page says Google-Agent can "perform actions". Make sure anything that changes data, such as orders or bookings, has the checks you'd want for any customer.
- Security tools. Some bot rules block anything that looks automated. If you want Google-Agent through, allow requests that match the token and come from Google's IP list.
If you want agents to complete purchases in Google's surfaces through a protocol rather than by clicking through pages, see how to sell in Google AI Mode and Gemini.
How Sill helps. Sill shows you which AI agents and bots visit your site and the pages they read. It verifies signed agent requests (Web Bot Auth) on Cloudflare and WordPress installs. Only agents that sign their requests can be verified this way. When an AI agent places an order, your rules approve it, decline it, or hold it for a person to review, and every decision is signed and recorded. To see how ready your site is for agents, run the free scan.
Questions people ask
Is Google-Agent the same as Googlebot?
No. Googlebot crawls automatically to build Google's search indexes and obeys robots.txt. Google-Agent visits because a user asked a Google-hosted agent to do something, and Google lists it with the user-triggered fetchers instead.
Does blocking Google-Agent hurt my Google rankings?
Google's documentation does not say it does. Google ties Search crawling preferences to the Googlebot token, and Google-Agent is documented separately as a user-triggered fetcher.
Which Google product uses Google-Agent?
Google's fetcher list says only that it is used by agents hosted on Google infrastructure to navigate the web and perform actions on user request. It does not name specific products, so we do not either.
Why does the Google-Agent user agent say Chrome?
Google's published strings include a Chrome version, shown as W.X.Y.Z, alongside the Google-Agent token. Look for the Google-Agent token itself, then verify the IP address, rather than trusting the string.
Sources
- Google: Google's user-triggered fetchers · checked 16 Nov 2026
- Google: Authenticate requests with Web Bot Auth · checked 16 Nov 2026
- Google: Verify requests from Google's crawlers and fetchers · checked 16 Nov 2026
- Google: Common crawlers · checked 16 Nov 2026
- Google: user-triggered-agents.json IP ranges · checked 16 Nov 2026