Agent Browser Integration

Pass

Audited by Gen Agent Trust Hub on Apr 3, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches and executes the agent-browser tool from Vercel's official package registry via bun x. This is a well-known service and the operation is standard for browser automation.
  • [COMMAND_EXECUTION]: Employs shell commands to interact with the agent-browser CLI and manage local screenshot storage. Commands are focused on project-specific UI verification and occur within the authorized project workspace.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external URLs during the browsing process, which is a standard functional requirement for UI verification tools.
  • Ingestion points: Target URLs provided in task context or configuration (SKILL.md).
  • Boundary markers: None explicitly defined for delimiters around web content.
  • Capability inventory: Subprocess execution via bun x agent-browser and file writes to repository-visible paths.
  • Sanitization: Relies on the underlying CLI tool's handling of web content; no additional sanitization logic is specified in the skill instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 3, 2026, 11:53 AM
Security Audit — agent-trust-hub — Agent Browser Integration