The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the lark-cli tool to interact with the Lark Open Platform. All commands and API interactions are consistent with the documented purpose of the skill and utilize standard parameters.
[DATA_EXFILTRATION]: The skill provides tools for downloading images and files. It includes explicit security instructions in lark-im-messages-resources-download.md to prevent path traversal by restricting output to relative paths and strictly forbidding the use of '..' in file paths.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it processes data from external chat messages. However, the risk is mitigated through design:
Ingestion points: External message content is retrieved via the +chat-messages-list, +threads-messages-list, and +messages-search shortcuts.
Boundary markers: The instructions encourage the use of structured JSON output (--format json) to help the agent distinguish between data and control instructions.
Capability inventory: The skill can send messages (+messages-send), reply to messages (+messages-reply), and download files (+messages-resources-download).
Sanitization: Mandatory 'Safety Constraints' are included in the instructions, requiring the agent to obtain explicit user approval for recipient, content, and identity before executing any send or reply actions.

lark-im