The Agent Skills Directory

[COMMAND_EXECUTION]: The skill operates by executing the lark-cli binary for all Lark platform interactions. It also runs a local bash script scripts/helper.sh for environment detection and profile management (using the macOS keychain for secure secret storage). Additionally, it generates and executes Node.js scripts to calculate complex layouts for whiteboard diagrams as seen in various scene templates.
[EXTERNAL_DOWNLOADS]: The skill utilizes npx to run the official @larksuite/whiteboard-cli package for rendering diagrams and fetches API specification data from trusted developer domains such as open.feishu.cn.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it reads data from messages, emails, and documents. However, it provides structured guidelines for the agent to verify user intent before performing actions and uses official CLI tools which act as a layer of separation from the ingested data.

better-lark