Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed ElevenLabs text-to-speech guide that uses expected audio generation, conversion, and messaging steps, with only minor operational cautions.
Before installing, be aware that text you synthesize may be sent to ElevenLabs and voice notes may be sent through WhatsApp when you invoke that workflow. Verify generated file paths before copying or deleting files, and only send voice messages to intended recipients.
Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
5. Cleanup:
rm /root/.openclaw/workspace/voice-temp.ogg
WhatsApp requires a non-empty message body to send voice notes. Use a single space as the message.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
**Finding more voices:**
- Browse: https://elevenlabs.io/voice-library
- v3-optimized collection: https://elevenlabs.io/app/voice-library/collections/aF6JALq9R6tXwCczjhKH
- API: `GET https://api.elevenlabs.io/v1/voices`
**Voice selection tips:**
- Use IVC (Instant Voice Clone) or premade voices - PVC not optimized for v3 yet
The skill documentation instructs the agent to use broad shell access to enumerate files in temporary directories, copy artifacts into the workspace, and remove them afterward. For a TTS skill, this expands behavior beyond narrowly scoped text-to-speech generation and creates opportunity for path confusion, unintended file access, or misuse of exec against files outside the intended output path.
This markdown instructs the user/agent to run an rm command to delete a workspace file, but the surrounding text does not include any safety warning beyond calling it 'Cleanup'. Under the markdown criteria for missing user warnings, behaviors that affect user data or system integrity should be disclosed clearly.
The document presents its primary correctness example using Hebrew text while the rest of the guide is in English, with no indication that users may choose their preferred language or that the example is region-specific. This can be interpreted as an unexplained language/locale preference in natural-language guidance.