The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes collaborator information and has the capability to modify document permissions. This creates a surface where malicious identifiers (like spoofed emails or names containing instructions) could potentially influence the agent's behavior during permission management tasks.
Ingestion points: Data returned from the list action in feishu_perm.
Capability inventory: Modifying document access levels via add and remove actions.
Boundary markers: None present in the skill definition.
Sanitization: Not described in the provided configuration.

feishu-perm