Update to NewsTune Studio

{
  "name": "newstune-voice-studio",
  "description": "Use when a user asks to browse or preview NewsTune podcast voices, choose one or two host voices, clone a voice through NewsTune's consent flow, or open NewsTune setup for a missing voice binding. Uses OAuth tools and keeps recording, consent, and uploads inside a secure NewsTune handoff.",
  "included_files": [
    {
      "relative_path": "references/voice-workflows.md",
      "size_in_bytes": 1871
    }
  ],
  "skill_md_contents": "---\nname: newstune-voice-studio\ndescription: \"Use when a user asks to browse or preview NewsTune podcast voices, choose one or two host voices, clone a voice through NewsTune's consent flow, or open NewsTune setup for a missing voice binding. Uses OAuth tools and keeps recording, consent, and uploads inside a secure NewsTune handoff.\"\n---\n\n# NewsTune Voice Studio\n\nUse `list_podcast_voices` and `prepare_voice_action` for voice work. You may also use paginated `list_podcast_series` and `get_podcast_series` only to resolve and verify a user-named owned series before creating its handoff. The plugin is OAuth-connected; never request an API key, browser token, password, raw recording, or voice sample in ChatGPT or Codex.\n\nTypical triggers include “help me choose a podcast voice,” “preview my voices,” and “clone my voice for this show.”\n\n## Workflow\n\n1. Clarify language, accent, host role, tone, pace, and solo or multi-host format.\n2. Call `list_podcast_voices`. Prefer an already accessible voice with a `previewUrl`; sharing that preview does not render new audio or spend credits.\n3. If the user names a series, follow `list_podcast_series.nextOffset` until the unique match is found and use `get_podcast_series` when verification is needed. Then call `prepare_voice_action` with `select_voice`, `selectionCount: 1` for solo or `selectionCount: 2` for two-host format, and that discovered `seriesId`; never invent an ID from its title. If no series is named, omit `seriesId`.\n4. Selecting an already accessible public/community voice is not voice cloning. Even when a voice is named after a celebrity or public figure, allow the user to preview and select it. Give a short reminder to avoid misleading endorsement, deceptive impersonation, or other unlawful use; do not block selection solely because of the identity or name.\n5. For cloning, ask the user to confirm that they understand they are responsible for having the appropriate rights or permission. Call the matching clone action only after that acknowledgement, then let NewsTune handle recording/upload and consent in the handoff.\n6. Explain that recording, upload, consent review, and any supported selection or binding happen inside NewsTune. The handoff only opens that flow; it neither performs nor proves that cloning, selection, or binding completed.\n\nRead [references/voice-workflows.md](references/voice-workflows.md) for the decision boundaries.\n\n## Refuse or stop when\n\n- The user states an intent to deceive, defraud, harass, bypass NewsTune consent, or falsely present the voice as an endorsement by the named person.\n- The handoff expires or NewsTune rejects authorization; create a fresh handoff only after the user asks.\n\nDo not infer endorsement from a voice name or availability. Never transfer raw audio through the model, and never claim a voice was selected, bound, or cloned until NewsTune reports completion.\n"
}