local-browser

Pass

Audited by Gen Agent Trust Hub on Jun 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Downloads and executes the @vm0/cli package from the npm registry using the npx utility. This package is managed by the same organization as the skill author.\n- [COMMAND_EXECUTION]: Uses shell commands to interact with the local browser environment and extension for tab management and page control.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted web content through snapshot and selection features, creating an attack surface for indirect instructions.\n
  • Ingestion points: External web content is read via page snapshot and page selection commands (SKILL.md).\n
  • Boundary markers: There are no explicit markers or instructions to isolate or ignore potentially malicious instructions embedded in web pages.\n
  • Capability inventory: The agent has capabilities to navigate, click, and type in the browser (SKILL.md).\n
  • Sanitization: No sanitization or validation procedures for the ingested external content are described.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 14, 2026, 01:52 AM
Security Audit — agent-trust-hub — local-browser