openclaw-voice-call-realtime

Warn

Audited by Gen Agent Trust Hub on Jul 8, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCREDENTIALS_UNSAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill's installation instructions require cloning a repository from a non-standard GitHub account (github.com/TristanBrotherton/openclaw-voice-call-realtime.git).
  • [REMOTE_CODE_EXECUTION]: The skill involves executing code downloaded from an external source via npm install and subsequent execution within the OpenClaw plugin environment.
  • [CREDENTIALS_UNSAFE]: The configuration requires sensitive credentials, specifically TWILIO_ACCOUNT_SID, TWILIO_AUTH_TOKEN, and OPENAI_API_KEY, to be stored in environment variables or configuration files (~/.openclaw/openclaw.json).
  • [COMMAND_EXECUTION]: The installation and setup process involves executing several shell commands, including git clone, npm install, and cloudflared for tunnel creation.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted real-time audio from phone calls, which is transcribed and fed back into the agent's context without explicit boundary markers or sanitization documented.
  • Ingestion points: Transcribed audio from inbound or outbound phone calls.
  • Boundary markers: Absent.
  • Capability inventory: Initiates calls, presses DTMF keys, and manages call outcomes.
  • Sanitization: None mentioned.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 8, 2026, 08:18 AM
Security Audit — agent-trust-hub — openclaw-voice-call-realtime