requirements-analysis-en

Pass

Audited by Gen Agent Trust Hub on Mar 14, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is designed to process untrusted requirements documents provided by users, which introduces an attack surface for indirect prompt injection.\n
  • Ingestion points: Requirements documents and user stories attached or provided as text inputs to the assistant, as defined in prompts/requirements-analysis_EN.md.\n
  • Boundary markers: The instructions do not specify explicit delimiters or "ignore embedded instructions" warnings for the requirements processing step.\n
  • Capability inventory: The skill includes utility scripts in the scripts/ directory for parsing and converting reports into various formats (JSON, CSV, DOCX, etc.).\n
  • Sanitization: Input text is processed based on its content structure without specific sanitization against malicious instructions embedded in the requirements document.\n- [COMMAND_EXECUTION]: The skill includes Python scripts that perform local command execution to manage file format conversions.\n
  • Execution method: The script scripts/batch_convert_templates.py and its wrappers use the subprocess.call method to invoke internal Python scripts like scripts/convert_formats.py using the system's Python interpreter.\n
  • Scope: Execution is restricted to the skill's own scripts and local files for the purpose of generating artifacts, with no network-based or arbitrary external command execution detected.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 14, 2026, 06:20 AM