Gardening Skills Wiki

Pass

Audited by Gen Agent Trust Hub on Feb 20, 2026

Risk Level: SAFE
Full Analysis
  • [Data Exposure & Exfiltration] (SAFE): The scripts read from and analyze files located in the user's local skills directory (~/.claude/skills). There are no commands that transmit data over a network or access sensitive system files outside the intended scope.
  • [Remote Code Execution] (SAFE): All scripts are local to the skill. There are no patterns involving downloading and executing remote code (e.g., curl | bash). The only external tool dependency is 'jq', which is a standard JSON processor.
  • [Privilege Escalation] (SAFE): The scripts do not use 'sudo', 'chmod 777', or modify system-level configurations. Operations are restricted to the user's own files.
  • [Obfuscation] (SAFE): The code is provided as plain-text bash scripts. No Base64 encoding, zero-width characters, or other obfuscation techniques were found.
  • [Indirect Prompt Injection] (SAFE): Although the scripts process markdown files that could contain untrusted data, they do so solely to generate a health report. The content of these files is searched for patterns (via grep/sed) but is never executed or used to construct complex prompts for the LLM that could trigger unintended actions.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 20, 2026, 08:51 AM