zhouyi-divination
Warn
Audited by Gen Agent Trust Hub on Jul 16, 2026
Risk Level: MEDIUMDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
- [DATA_EXPOSURE]: The skill instructs the agent to read a sensitive file containing personal information (八字/Bazi) from a hardcoded absolute path:
/Users/taisenzhuang/.openclaw/workspace/zhouyi-divination/mingzhu.md. Accessing files via absolute paths in user-specific directories is a security concern as it targets private data structures. - [COMMAND_EXECUTION]: The skill mandates the automatic creation of folders and files in the user's iCloud directory (
~/Library/Mobile Documents/com~apple~CloudDocs/Documents/OpenClaw/扁告/运势/) without requiring per-instance user confirmation. This represents an automated file system write operation. - [DATA_EXFILTRATION]: By forcing the storage of divination results—which are derived from the user's sensitive personal data—into a cloud-synced directory (iCloud), the skill ensures that private information is synchronized to external servers.
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from
USER.md, which is an external and potentially untrusted input source. - Ingestion points:
USER.md,/Users/taisenzhuang/.openclaw/workspace/zhouyi-divination/mingzhu.md. - Boundary markers: None identified.
- Capability inventory: File read (personal data), File write (iCloud directory).
- Sanitization: No evidence of input validation or sanitization before processing user data.
Audit Metadata