lark-whiteboard-cli
Warn
Audited by Gen Agent Trust Hub on Apr 10, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on shell commands including
node,npx, andlark-clito perform diagram rendering, calculations, and uploads to Lark Suite. - [COMMAND_EXECUTION]: It specifically directs the agent to generate Node.js source files (e.g.,
diagram.gen.js) and execute them usingnodeto handle the complex mathematical positioning required for certain diagram types like fishbone or flywheel charts. - [REMOTE_CODE_EXECUTION]: The workflow for generating these Node.js scripts is vulnerable to code injection. Templates provided in files such as
scenes/flywheel.mdandscenes/fishbone.mduse simple string interpolation for user-provided labels (e.g.,{{CENTER_TITLE}}). If a user provides a payload designed to escape the JavaScript string literal context, they can achieve arbitrary code execution within the Node.js process during the calculation phase. - [EXTERNAL_DOWNLOADS]: The skill fetches and executes the
@larksuite/whiteboard-clitool vianpx. This is an official tool provided by the author (larksuite) and is recognized as a legitimate vendor resource.
Audit Metadata