watch

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt explicitly tells the agent to AskUserQuestion for Groq/OpenAI API keys and then write lines like GROQ_API_KEY=... or OPENAI_API_KEY=... into ~/.config/watch/.env, which requires the LLM to accept and embed secret values verbatim (high exfiltration risk).

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

• Third-party content exposure detected (high risk: 0.85). Outsider-authored free text can enter the LLM context via runtime parsing of externally fetched WebVTT captions (downloaded from the user-supplied URL by scripts/download.py and read/converted to transcript text by scripts/transcribe.py, then printed in scripts/watch.py under “## Transcript” and included in the agent’s subsequent context).