installer

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADS
Full Analysis
  • Persistence Mechanisms (MEDIUM): The skill creates 'hook scripts' for tracking skill invocations. This introduces a persistence mechanism where custom logic can be executed every time the agent performs specific tasks.
  • Data Exposure & Exfiltration (MEDIUM): The skill interacts with sensitive file paths in the agent's private directory (~/.claude/activity-logs/). Accessing and writing to internal agent metadata storage is a high-risk behavior that could lead to the unauthorized collection of usage patterns.
  • Unverifiable Dependencies & Remote Code Execution (LOW): The skill's primary logic is delegated to external files in a references/ directory. This indirect execution model obscures the specific commands being run by the agent during the installation process.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 06:25 PM