integrated-browser
Pass
Audited by Gen Agent Trust Hub on Jul 1, 2026
Risk Level: SAFE
Full Analysis
- Process Isolation Architecture: The skill describes a multi-process architecture (Main, Shared, and Renderer) that separates UI rendering from high-privilege operations like native view creation and automation. This design limits the potential impact of a compromise in the renderer process.
- Indirect Prompt Injection Awareness: The documentation explicitly identifies web page content as untrusted input. It provides guidance on treating such content as a potential vector for prompt injection, encouraging the use of privacy gates and availability checks to mitigate risks when agents interact with browser content.
- Secure Session Management: The architecture utilizes dedicated sessions to manage cookies, storage, and sensitive permissions such as access to local files. These are implemented as central trust boundaries to ensure that security-sensitive capabilities are properly gated.
- Controlled Automation Access: The skill details how browser automation (via Playwright and CDP) is managed through proxies and privacy gates rather than providing direct, unrestricted access to the browser core. This ensures that agent actions are subject to user-defined visibility and security constraints.
Audit Metadata