The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill enables an AI agent to perform sensitive permission management actions (add, remove, list collaborators) on the Feishu platform. This functionality is triggered by natural language mentions of sharing or permissions, creating a potential attack surface where malicious instructions embedded in processed data could influence the agent's actions.
Ingestion points: The skill is activated by user mentions of sharing or permissions in chat or document contexts (SKILL.md).
Boundary markers: The instructions lack explicit boundary markers or directives to distinguish between legitimate user commands and malicious instructions that might be present in document content being processed by the agent.
Capability inventory: The feishu_perm tool allows for adding collaborators with full_access permissions and removing existing collaborators, providing significant control over document access control lists.
Sanitization: There is no evidence of sanitization or human-in-the-loop verification requirements for the inputs that trigger these sensitive operations, although the skill documentation correctly notes that the tool is disabled by default.

feishu-perm