BrowserSkill
An open-source AI browser automation tool from Tencent. By reusing the user's existing authenticated browser sessions, it allows AI agents to operate pages requiring authentication without credentials. It adopts a local daemon architecture to ensure security and auditability.
Timeline (last 90 days)
Tencent has open-sourced the BrowserSkill project under the MIT license.
BrowserSkill lets AI directly borrow the user's already logged-in browser session without requiring an account username or password.
BrowserSkill provides AI with a command-line tool and a browser extension.
In BrowserSkill, AI opens a separate window to perform operations, without occupying the tab the user is currently browsing.
BrowserSkill places a local daemon process between the AI and the browser, through which all operations pass and leave traceable, auditable records.
When encountering steps that AI cannot handle, such as CAPTCHAs, BrowserSkill pauses and hands control over to a human.
The capabilities supported by BrowserSkill include post-login page scraping, internal system form filling, full-page long screenshots, and online request troubleshooting.
BrowserSkill supports Chrome and Edge browsers, covering three major operating systems: Windows, macOS, and Linux.
BrowserSkill can be directly accessed by agents that can use the command line, such as coding agents like Cursor and Claude Code.
BrowserSkill only supports Chromium-based browsers and does not support Firefox or Safari.
2 more timeline events
All Facts (12)
Tencent has open-sourced the BrowserSkill project under the MIT license.
50%UnverifiedBrowserSkill lets AI directly borrow the user's already logged-in browser session without requiring an account username or password.
50%UnverifiedBrowserSkill provides AI with a command-line tool and a browser extension.
50%UnverifiedIn BrowserSkill, AI opens a separate window to perform operations, without occupying the tab the user is currently browsing.
50%UnverifiedBrowserSkill places a local daemon process between the AI and the browser, through which all operations pass and leave traceable, auditable records.
50%UnverifiedWhen encountering steps that AI cannot handle, such as CAPTCHAs, BrowserSkill pauses and hands control over to a human.
50%UnverifiedThe capabilities supported by BrowserSkill include post-login page scraping, internal system form filling, full-page long screenshots, and online request troubleshooting.
50%UnverifiedBrowserSkill supports Chrome and Edge browsers, covering three major operating systems: Windows, macOS, and Linux.
50%UnverifiedBrowserSkill can be directly accessed by agents that can use the command line, such as coding agents like Cursor and Claude Code.
50%UnverifiedBrowserSkill only supports Chromium-based browsers and does not support Firefox or Safari.
50%UnverifiedBrowserSkill's extension requires a one-time manual installation and is not ready to use out of the box.
50%UnverifiedOne BrowserSkill extension can only serve one agent, which limits multi-agent concurrent scenarios.
50%