agent-browser

Pass

Audited by Gen Agent Trust Hub on Apr 30, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references and downloads configuration from Vercel Labs' official GitHub repository. As this is a trusted organization, the references are considered safe.
  • [COMMAND_EXECUTION]: Utilizes the Bash tool to execute CLI commands for browser automation, including installation and runtime interactions.
  • [PROMPT_INJECTION]: Identifies the potential for indirect prompt injection given that the skill processes untrusted content from the web. The skill provides proactive hardening measures to mitigate this risk.
  • Ingestion points: Web content is ingested via browser snapshots and navigation as described in SKILL.md and references/snapshot-refs.md.
  • Boundary markers: The skill documents the use of AGENT_BROWSER_CONTENT_BOUNDARIES to isolate external content from the agent's instructions.
  • Capability inventory: The skill uses the Bash tool to run interaction commands like click, fill, and eval within the browser sessions.
  • Sanitization: Recommends the use of action policies and domain allowlists to restrict the browser's execution environment.
  • [DATA_EXFILTRATION]: Provides instructions for managing authentication states through session persistence. It adheres to security best practices by recommending environment variables for credential management.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 30, 2026, 05:51 AM
Security Audit — agent-trust-hub — agent-browser