speech-pathology-ai

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

• Third-party content exposure detected (high risk: 0.70). The skill explicitly ingests and processes arbitrary user-provided audio and transcriptions at runtime (e.g., MellifluoFeedbackEngine.process_audio_stream calls ChildrenSpeechRecognizer.transcribe_with_confidence and then scores/transforms those transcripts), so it reads untrusted user-generated content as part of its workflow.