dingtalk-openclaw-connector
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the installation of an external package
@dingtalk-real-ai/dingtalk-connectorfrom the NPM registry to function. - [REMOTE_CODE_EXECUTION]: The installation process involves running
npx -y @dingtalk-real-ai/dingtalk-connector install, which executes code from the external package on the host machine. - [INDIRECT_PROMPT_INJECTION]: The skill processes message content from DingTalk users, creating a vulnerability where malicious instructions could be embedded in data processed by the agent.
- Ingestion points: Messages received via the
MessageContextin thehandleMessagefunction (SKILL.md). - Boundary markers: Absent; the skill does not use delimiters or instructions to ignore commands within the message text.
- Capability inventory: Significant capabilities are exposed, including messaging (
sendMessage), document management (createDocument,appendDocument), and task scheduling (createTodo,createCalendarEvent). - Sanitization: The documentation does not describe any sanitization or validation of incoming message text before it is used by the AI agent.
Audit Metadata