The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to install a global Node.js package (@qingchencloud/openclaw-zh) and manage additional plugins via openclaw skills install. These resources originate from third-party sources that are not part of the verified trusted vendor list.
[DATA_EXPOSURE]: The documentation encourages the use of an unverified third-party API platform (https://gpt.qt.cool/v1) as an LLM provider. Sending sensitive API keys (OpenAI/Claude) through a third-party bridge or proxy service introduces a risk of credential interception or data exposure.
[PERSISTENCE]: The skill includes functionality to install the application as a system-level daemon (openclaw gateway install), which ensures the process remains active across sessions and automatically restarts during the system boot sequence.
[PRIVILEGE_ESCALATION]: Several commands, including system service installation and global NPM package management, typically require administrative or root privileges. The skill onboarding process guides the user through these high-privilege operations.
[INDIRECT_PROMPT_INJECTION]: The skill integrates with external messaging services (WhatsApp, Telegram, Discord), creating a surface for processing untrusted data.
Ingestion points: Incoming messages from configured chat platform channels.
Boundary markers: None specified; the skill instructions do not define delimiters to separate user input from system prompts.
Capability inventory: The agent has access to the local file system (~/.openclaw), network operations, and the ability to install further software extensions.
Sanitization: The skill does not document any mechanisms for filtering or sanitizing external platform content before processing.
[DYNAMIC_EXECUTION]: The openclaw skills install and openclaw update commands facilitate the runtime downloading and loading of executable code and extensions, which are not statically defined in the skill itself.

openclaw-chinese-ai-assistant