Browser Agent
A category of AI agent capable of autonomously controlling a web browser to complete complex web-based tasks. These agents operate on a perception-planning-execution loop, enabling them to interpret page content, plan sequences of actions, and perform interactions such as clicking and text input.
Timeline (last 90 days)
The rule the author set for the browser agent is that it may copy and paste, but may not read, recite, narrate, or transcribe the key.
The governance approach based on credential escrow cannot cover browser agents, because they perform pixel-level click operations under a session where the user is already logged in and there are no independent credentials that can be withheld by a gateway.
Browser agents typically adopt a Perception-Planning-Action loop architecture.
Completing a moderately complex browser agent task (such as filling out a form or extracting multi-page data) often requires dozens to hundreds of model calls.
Browser agents have six main implementation approaches: browser extensions, cloud-based Chromium, local CLI daemon processes, AI-native browsers, page-embedded scripts, and iframe nesting.
All Facts (5)
The rule the author set for the browser agent is that it may copy and paste, but may not read, recite, narrate, or transcribe the key.
50%UnverifiedThe governance approach based on credential escrow cannot cover browser agents, because they perform pixel-level click operations under a session where the user is already logged in and there are no independent credentials that can be withheld by a gateway.
50%UnverifiedBrowser agents typically adopt a Perception-Planning-Action loop architecture.
50%UnverifiedCompleting a moderately complex browser agent task (such as filling out a form or extracting multi-page data) often requires dozens to hundreds of model calls.
50%UnverifiedBrowser agents have six main implementation approaches: browser extensions, cloud-based Chromium, local CLI daemon processes, AI-native browsers, page-embedded scripts, and iframe nesting.
50%