peach-e2e-browse
Pass
Audited by Gen Agent Trust Hub on May 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill interacts with the local operating system to manage Chrome Beta instances using the Chrome DevTools Protocol (CDP). It executes shell commands to start the browser and uses automation CLIs to perform actions like page navigation, element clicking, and form filling. It heavily utilizes JavaScript execution within the browser context via
evalto extract data and manipulate the DOM. - [EXTERNAL_DOWNLOADS]: The documentation identifies standard, well-known automation tools (
agent-browser,@playwright/cli) available through the NPM registry for environment configuration. These are standard dependencies for the skill's stated purpose. - [PROMPT_INJECTION]: The skill is designed to ingest and process content from external websites, which constitutes an indirect prompt injection surface (Category 8). The skill effectively mitigates this risk by including explicit 'Force Gates' and 'Decision Trees' that instruct the agent to halt and seek user approval before proceeding with operations on sensitive domains or authentication-heavy sessions.
Audit Metadata