siyu-qunfa
Pass
Audited by Gen Agent Trust Hub on Aug 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill facilitates the execution of local Python scripts, specifically
qunfa_lint.pyfor compliance scanning andatoms_query.pyfor metadata retrieval. These executions are standard for the skill's operational logic and occur within the local environment. - [PROMPT_INJECTION]: The skill processes user-supplied marketing copy as input for its auditing tool, creating a surface for indirect prompt injection.
- Ingestion points: Marketing text input provided by the user for scanning (SKILL.md).
- Boundary markers: Absent in the shell pipeline used to pass content to the script.
- Capability inventory: Local script execution and filesystem access (scripts/qunfa_lint.py).
- Sanitization: The skill relies on the script's internal logic for analysis and does not sanitize the input before processing.
Audit Metadata