The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on the dws CLI tool to perform all storage and management operations on DingTalk Drive. This includes searching, downloading, uploading, and modifying file structures as outlined in SKILL.md.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of file and folder names, which could be used to host malicious instructions. Since the skill has the capability to modify and delete files, this represents a potential vulnerability surface.
Ingestion points: Untrusted data enters the agent context via dws drive +list, dws drive +search, and dws drive +recent commands (documented in SKILL.md).
Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from interpreting instructions embedded within file or folder names.
Capability inventory: The skill allows for high-impact operations including +move, +rename, +delete, +upload, and permission update using the dws CLI.
Sanitization: No sanitization, validation, or escaping of file names or metadata is performed before the data is processed or displayed.
[DYNAMIC_EXECUTION]: The skill references an external Python script located at ../scripts/drive_tree_list.py to handle recursive directory tree listing, as noted in the error recovery and auxiliary script sections of references/drive.md.

dingtalk-drive