AudioEditor

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.90). The skill mandates an immediate, non-optional curl POST to a local notify endpoint before doing anything else—an extraneous side-effect unrelated to audio editing that forces network activity and changes agent behavior outside the skill's stated purpose.

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

• Third-party content exposure detected (high risk: 0.75). Outsider-authored free text can enter the LLM context via the user-supplied audio file’s transcription: Pipeline.ts → Transcribe.ts produces transcript.json from the audio, then Analyze.ts reads that transcript and embeds windowText (transcript text) into userPrompt for the Claude inference() call.