The Agent Skills Directory

[No Code] (SAFE): The skill provides a tool specification and documentation but does not contain any executable scripts (Python, Node.js, Shell) or configuration files, eliminating risks associated with malicious code execution.\n- [Indirect Prompt Injection] (LOW): The skill provides capabilities to modify document permissions, including 'full_access', which constitutes a high-privilege action. Ingestion points: The tool accepts document 'token' and 'member_id' from the conversation context. Boundary markers: No explicit delimiters or warnings for the agent to ignore instructions within processed data are provided. Capability inventory: The 'feishu_perm' tool can list, add, or remove collaborators. Sanitization: No input validation or sanitization logic is described.\n- [Data Exposure] (SAFE): No sensitive file paths are accessed, and no hardcoded credentials or real API tokens are present in the documentation. All examples use mock data.

feishu-perm