create-project

Pass

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: SAFE
Full Analysis
  • [DATA_EXFILTRATION]: The skill reads and writes project data to local CSV files specified by the $PM_PATH environment variable. No network-based exfiltration or access to sensitive system files (e.g., credentials, SSH keys) was detected.
  • [COMMAND_EXECUTION]: The skill uses Python scripts to automate project creation. These scripts rely on standard, well-known libraries such as pandas, uuid, and datetime. The logic is limited to data manipulation and file system persistence of project records.
  • [PROMPT_INJECTION]: The skill accepts user-provided input for project fields like names and descriptions. While these values are stored and could potentially be re-processed by an agent, the current implementation provides a standard data-entry surface without instructions targeting the model's safety constraints or system prompt.
Audit Metadata
Risk Level
SAFE
Analyzed
May 20, 2026, 09:22 AM
Security Audit — agent-trust-hub — create-project