Security audit
Security checks for vulnerabilities and agentic risk
The skill is coherent and not deceptive, but it needs review because ambient speech can trigger real OpenClaw actions through broad default wake phrases.
Install only if you are comfortable with an always-listening voice workflow that can send commands to your agent. Before enabling it, replace common wake words with a unique phrase, approve only trusted speakers, and require confirmation for email, text, calendar, credential-related notes, and custom OpenClaw CLI actions.
The description emphasizes convenience and capabilities but does not clearly warn that ambient speech is monitored for wake words and that captured commands are forwarded for execution. This omission undermines informed consent and can cause users to deploy always-listening, action-taking functionality without understanding the privacy and security consequences.
The documented default wake words include very common phrases such as "take notes" and "send an email," which are highly prone to false activation in ambient speech. Because the skill captures follow-on speech and dispatches commands to an OpenClaw CLI, an accidental trigger could cause unauthorized or unintended agent actions, especially in shared environments.
The examples explicitly normalize broad activation phrases like "take notes," which can plausibly occur in ordinary conversation and trigger command capture unintentionally. In a voice-command skill that forwards detected commands to an agent for execution, accidental activation can lead to unintended actions or disclosure of spoken content.
No suspicious patterns detected.
