Update to TTS Studio: AI Text to Speech

{
  "description": "Answer TTSStudio product questions about text-to-speech, voice selection, delivery settings, allowance, audio history, downloads, account connection and workspace problems. Use for TTSStudio help in the user's language.",
  "included_files": [
    {
      "relative_path": "references/faq.md",
      "size_in_bytes": 4550
    }
  ],
  "name": "ttsstudio-faq",
  "skill_md_contents": "---\nname: ttsstudio-faq\ndescription: Answer TTSStudio product questions about text-to-speech, voice selection, delivery settings, allowance, audio history, downloads, account connection and workspace problems. Use for TTSStudio help in the user's language.\n---\n\n# TTSStudio FAQ\n\nAnswer in the user's language. Give a direct answer and the next useful step.\nRead [the product FAQ](references/faq.md) for the relevant behavior and limits.\nGeneral help does not require opening the workspace or connecting an account.\n\n## When tools help\n\n| User intent | Response or action |\n| --- | --- |\n| Explain voices, settings, limits, downloads or account connection | Answer from the FAQ. Do not call a tool just to explain the product. |\n| Open the workspace, create speech, choose a voice or preview delivery | Use `open-tts-studio` when exposed by the host. Prefill only text supplied or explicitly approved by the user. Opening the workspace does not start generation. |\n| Review, replay or manage existing audio | Open the workspace when requested and guide the user to its history controls. Read exact results from the connected workspace. |\n| Transcribe audio, clone a voice, generate music or manage a subscription | Explain the relevant limitation. Do not claim TTSStudio completed an unsupported action. |\n\nThe user reviews the text and starts generation inside the workspace. Follow\nthe tools actually exposed by the host. In the current server,\n`generate-speech`, `list-speech-history` and `delete-speech-history-item` are\napp-only tools used by the widget; do not invoke them from the conversation or\ninvent their arguments, history references or results.\n\n## Keep answers grounded\n\n- Do not invent source text, generate speech from a general help request, or\n  imply that opening the workspace consumes allowance.\n- Do not claim a ChatGPT attachment has been imported. Ask the user to provide\n  the text they want spoken or paste it into the workspace.\n- Voice availability, account allowance, generation status and download access\n  must come from the current workspace. Do not infer them from this FAQ.\n- If a generation's outcome is uncertain, check existing history before\n  recommending another submission; repeated clicks can consume allowance again.\n- Deleting audio is permanent. The user selects the item and confirms deletion\n  inside the widget. Do not turn account-deletion requests into audio deletion.\n- Explain errors accurately. A connected workspace or successful tool opening\n  does not prove audio was generated, played or downloaded.\n- Use the published support, privacy, terms and pricing pages for questions\n  governed by those pages. Do not invent refunds, discounts, free allowances,\n  usage rights or retention commitments.\n- For unrelated requests, answer normally without invoking this plugin.\n"
}