openclaw-voice-call-realtime

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.75). The required runtime workflow involves generating call transcripts and summaries from the live phone conversation (host/IVR/caller speech), which is outsider-authored free text that can be fed back into the agent/LLM context via the “get_transcript”/transcript+summary path.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The install instructions clone and install code from https://github.com/TristanBrotherton/openclaw-voice-call-realtime.git which becomes an OpenClaw plugin executed by the agent and thus is a required remote code dependency that controls agent behavior.