transcription-speech-to-text-hebrew

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly says "If the user provides the API key directly in the chat, write it into textops_settings.json," which requires the LLM to accept and embed a secret value verbatim (high exfiltration risk).

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly downloads and ingests audio from open/public sources (e.g., YouTube via scripts/download_audio.py and arbitrary HTTP/HTTPS URLs accepted by scripts/transcribe.py), produces and reads transcripts of user-generated/untrusted audio as part of its workflow (including using transcript-derived speaker info to determine diarization), so untrusted third‑party content is consumed and can influence processing.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill's scripts/download_audio.py calls pip at runtime to install yt-dlp (via "python -m pip install yt-dlp"), which fetches and installs remote code from PyPI (e.g. https://pypi.org/project/yt-dlp) and then executes that tool to download YouTube audio — a runtime fetch that installs/executes external code and is required for YouTube download functionality.