AGENT LAB: SKILLS

claude-skill

Warn

Audited by Snyk on Feb 15, 2026

Risk Level: MEDIUM
Full Analysis

MEDIUM W013: Attempt to modify system services in skill instructions.

  • Attempt to modify system services in skill instructions detected (high risk: 0.90). The skill defaults to auto-accepting file edits, encourages fully autonomous execution, and explicitly documents a "bypassPermissions" mode that skips all permission prompts (and examples allowing Bash/tool use), which enables and encourages bypassing security prompts and making potentially privileged/system changes on the host.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 15, 2026, 09:04 PM