What is NovaAct? AI crawler guide

NovaAct by Amazon: Amazon Nova Act agent token reported for browser-style task execution. Check its reported user-agent, robots.txt behavior, source, and verification guidance.

Amazon Nova Act agent token reported for browser-style task execution.

What is NovaAct?

NovaAct is an interaction bot operated by Amazon. It is reported as an agent token associated with Amazon Nova Act, a browser-style task execution system. Unlike traditional crawlers that primarily index content, NovaAct performs agentic navigation, meaning it can interact with web pages on behalf of a user to complete tasks. The user-agent token is simply "NovaAct". Its behavior is not fully documented, and its robots.txt posture is unverified, so site owners should treat it as an agent that may execute actions rather than just fetch pages.

What it's for

For a site owner, a hit from NovaAct may indicate that an automated agent is navigating your site to perform tasks for a user, rather than just indexing content. This could affect how your site's interactive elements are used, potentially leading to unintended actions or resource consumption. Understanding this distinction helps in deciding whether to allow or block such agentic traffic based on your site's functionality and business needs.

Allowing NovaAct only creates the possibility of retrieval. To find out whether your pages are selected, measure which answer engines actually cite your pages using a stable query set.

How to handle NovaAct

To manage NovaAct, you can add a robots.txt rule targeting its user-agent token. Since its posture is unverified, blocking it may prevent agentic task execution on your site. Consider the nature of your site: if it relies on user-specific interactions that should not be automated, blocking could be beneficial. If you are unsure, monitor its activity before deciding.

robots.txt rule

User-agent: NovaAct Disallow: /

Blocking cost

Blocking NovaAct may prevent Amazon Nova Act from performing automated tasks on your site, which could reduce agent-driven traffic or functionality but is unlikely to affect traditional search indexing.

Examples

• An e-commerce site notices NovaAct in its logs, suggesting an agent may be adding items to a cart or checking out on behalf of a user.
• A web application sees NovaAct requests that mimic form submissions, indicating automated task completion rather than content scraping.
• A content site observes NovaAct navigating through interactive tutorials, potentially executing steps without human intervention.

Related bots

• GoogleAgent-Mariner: Also tracked as a live fetcher.
• AmazonBuyForMe: Another Amazon live fetcher to compare.
• ChatGPT Agent: Also tracked as a live fetcher.
• Claude-Code: Also tracked as a live fetcher.
• Gemini-Deep-Research: Also tracked as a live fetcher.
• Google-Gemini-CLI: Also tracked as a live fetcher.
• cohere-ai: Also tracked as a live fetcher.
• Meta-ExternalFetcher: Also tracked as a live fetcher.
• Claude-Web: Also tracked as a live fetcher.
• Robots.txt: Robots.txt is the control file used to allow or block NovaAct.
• Alexa: NovaAct gives crawler context for Alexa.