majia-siyu
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONPERSISTENCEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes several internal Python scripts and CLI tools to process tasks. For instance, it invokes modules/siyu-pyq/scripts/pyq_lint.py to perform compliance checks on marketing copy and siyu-plan to generate task execution plans. These scripts are self-contained within the skill package.
- [PERSISTENCE]: Consultation conclusions and client profiles are stored locally in the ~/.siyu/clients/ directory. This persistence mechanism is designed with a specific privacy workflow that requires the agent to mask sensitive information and obtain explicit user consent before writing any files to disk.
- [EXTERNAL_DOWNLOADS]: The siyu-update module facilitates synchronization with the official GitHub repository for the skill (maojiebc/majia-siyu-team). It uses standard tools like git and npx to fetch and install updates, ensuring the environment remains current with the author's official releases.
- [INDIRECT_PROMPT_INJECTION]: As the skill processes untrusted user-supplied marketing materials and business data, it presents a potential indirect injection surface. This risk is mitigated by explicit instructions to treat user data as non-executable information and by a mandatory compliance scanning step (using a static regex-based lexicon) that verifies all generated outputs for safety before they are finalized.
Audit Metadata