policy-sensitivity-brief

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to activate virtual environments and run Python scripts (references/data-queries.md).
  • [EXTERNAL_DOWNLOADS]: The skill performs automated package installation using pip install from a requirements file located in a sibling directory (references/data-queries.md).
  • [REMOTE_CODE_EXECUTION]: The skill relies on and executes code from a relative path outside its own directory structure (../findata-toolkit-cn/scripts/views_runner.py), which is an external code dependency provided by the skill author (references/data-queries.md).
  • [PROMPT_INJECTION]: The skill processes untrusted data from external sources such as government policy websites and news APIs (references/methodology.md).
  • Ingestion points: External news, policy data, and AKShare data (SKILL.md, references/methodology.md).
  • Boundary markers: None identified in the prompt instructions to isolate external data.
  • Capability inventory: Shell execution, file access, and Python execution (references/data-queries.md).
  • Sanitization: No explicit data sanitization or validation logic is defined in the documentation.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 03:23 AM