feishu-drive
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHPROMPT_INJECTIONCOMMAND_EXECUTIONNO_CODE
Full Analysis
- Indirect Prompt Injection (HIGH): The skill creates a vulnerability surface by combining external data ingestion with destructive capabilities.
- Ingestion points: The
listandinfoactions inSKILL.mdretrieve file names and metadata from Feishu Drive. - Boundary markers: Absent. There are no instructions or delimiters to prevent the agent from obeying instructions embedded in file names.
- Capability inventory: The skill allows
deleteandmoveactions which provide high-impact side effects on user data. - Sanitization: Absent. No validation or filtering of external strings is mentioned in the tool specification.
- Command Execution (LOW): While not executing shell commands, the tool enables significant state-changing operations (delete/move) within the Feishu environment.
- No Code (INFO): The skill consists of documentation and tool definitions without executable script files.
Recommendations
- AI detected serious security threats
Audit Metadata