What is agent-browser?
agent-browser converts the page accessibility tree into a snapshot with references, enabling the Agent to locate elements and perform subsequent actions. The official website currently introduces a native Rust cross-platform application that supports macOS, Linux, and Windows, and provides documentation for features such as sessions, authentication status, storage, and networking.
Observability features include recording, live streaming, debugging, performance monitoring, and diff tools, making it easy to inspect automation workflows. Browser installation, session permissions, and model services are separate components; the installation command shown on the page is merely part of the product documentation and was not actually executed in this instance. Furthermore, compatibility with a particular Agent should not be interpreted as official authorization for all web‑page operations.
Key features
Compact Text Snapshot
Reference-style element selection
Navigation, Forms, and Screenshots
Session and Authentication Status
Network and Debugging Control
Recording, Streaming, and Differential Observation
Use cases and audience
Agent and browser automation developer
- Provide the Agent with inspectable browser actions.
- Test the authorization web page flow
- Logging and troubleshooting automation tasks
Authorization web pages, browser status, and operation commands
Snapshots, screenshots, web page tasks, and debug logs
How to use
- 01
Verify the system and browser requirements according to the official documentation.
- 02
Obtain a snapshot within the authorized scope, then locate the desired element.
- 03
Inspect operations and logs, and manage sessions as required by the task.
Official website & sources
Updated on 2026-10-25
