agent-browser

Pass

Audited by Gen Agent Trust Hub on Mar 12, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires the global installation of the agent-browser package from NPM and the subsequent download of the Chromium browser via the agent-browser install command. These are functional requirements for the CLI tool.
  • [COMMAND_EXECUTION]: The skill relies on executing bash commands (e.g., agent-browser open, agent-browser click) to drive browser automation.
  • [PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection by processing external web content.
  • Ingestion points: Untrusted data enters the agent context through the agent-browser snapshot, get text, and get html commands described in SKILL.md.
  • Boundary markers: The instructions do not define clear delimiters or provide the agent with specific warnings to ignore instructions found within the scraped web content.
  • Capability inventory: The skill provides extensive capabilities to interact with the system and web, including form filling, element interaction, and navigation.
  • Sanitization: There is no mechanism described for sanitizing or filtering retrieved HTML or text content before the agent processes it.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 12, 2026, 03:22 AM