browser
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: In
scripts/start.cjs, the script spawns a local Google Chrome instance by executing a platform-specific binary path (child_process.spawn) with arguments configured to enable remote debugging parameters. - [DYNAMIC_EXECUTION]: In
scripts/eval.cjs, user-supplied arguments are dynamically evaluated inside the browser via the Chrome DevTools Protocol (Runtime.evaluate). While this is the core feature of the automation helper, executing unvalidated code within a browser space can result in code execution hooks or data leakage context if untrusted parameters flow into this module. - [INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection via
scripts/pick.cjsandscripts/nav.cjs. If an agent uses this tool to navigate to an external website controlled by an attacker or parse raw elements, instructions embedded inside that web content could compromise subsequent workflow actions.
Audit Metadata