agent-browser
Pass
Audited by Gen Agent Trust Hub on Jun 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides templates and examples (in Python and Bash) for wrapping and executing the 'agent-browser' CLI tool. This includes using subprocess calls to interact with the browser daemon.
- [EXTERNAL_DOWNLOADS]: The documentation includes instructions for installing the automation tool via NPM and downloading skill metadata directly from Vercel Labs' official GitHub repository.
- [PROMPT_INJECTION]: As a tool designed to interact with external websites and process their accessibility trees (snapshots), the skill exposes an attack surface for indirect prompt injection if the agent processes content from untrusted web sources.
- [COMMAND_EXECUTION]: The skill documentation describes how to install necessary system dependencies on Linux using package managers (apt, dnf), which is a standard requirement for browser automation frameworks like Playwright.
Audit Metadata