NYC

get-available-resources

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • Indirect Prompt Injection (LOW): The skill generates strategic recommendations in a JSON file that the agent is prompted to follow, creating a multi-step influence chain (Category 8b). \n
  • Ingestion points: .claude_resources.json file generated at runtime. \n
  • Boundary markers: Absent; instructions direct the agent to 'Read and Apply Recommendations'. \n
  • Capability inventory: Local subprocess execution (python, system tools) and file system write access. \n
  • Sanitization: Absent; output is generated directly from system utility results. \n- External Downloads (LOW): The skill requires the psutil library. While standard, it is an external package dependency not included in the 'Trusted Sources' list. \n- Command Execution (LOW): Uses subprocesses to run a local discovery script and system-level hardware utilities (e.g., nvidia-smi, rocm-smi).
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:31 PM