Update to YouCam

{
  "description": "Use when the user wants to edit, retouch, enhance, restore, transform, or otherwise modify an existing image or photo, including background or object changes. Do not use when the primary request is appearance analysis, style advice, or creating a new image from scratch.",
  "included_files": [
    {
      "relative_path": "agents/openai.yaml",
      "size_in_bytes": 504
    },
    {
      "relative_path": "assets/icon.png",
      "size_in_bytes": 42176
    },
    {
      "relative_path": "references/mcp-errors.md",
      "size_in_bytes": 1557
    }
  ],
  "name": "edit-image",
  "skill_md_contents": "---\nname: edit-image\ndescription: Use when the user wants to edit, retouch, enhance, restore, transform, or otherwise modify an existing image or photo, including background or object changes. Do not use when the primary request is appearance analysis, style advice, or creating a new image from scratch.\n---\n\n# Edit Image\n\nDelegate matching requests to YouCam. Preserve the user's intent and let the\nspecialist and its MCP App complete the beauty or creative-image task.\n\n## Safety boundary\n\nDo not call YouCam for clinical dermatology or medical concerns such as a\nsuspected infection, severe reaction, diagnosis, treatment, or medication\ninteraction. Recommend an appropriate clinician and provide only clearly\nlabeled, non-medical context.\n\n## Call YouCam\n\nCall `create-beauty-content` for each matching request:\n\n- Set `source_skill` to `\"edit-image\"` on every call. This fixed value\n  identifies the loaded skill for QA tracing; never infer or change it from the\n  user's wording.\n- Put the user's request in `user_message`, adding only the minimum recent\n  context needed to make it understandable.\n- Omit optional fields instead of passing `null`.\n- Omit `session_id` on the first call.\n- Do not broaden the request or add goals the user did not ask for.\n- Do not issue concurrent `create-beauty-content` calls in the same conversation.\n\n## Pass images\n\nNever read an attachment with shell commands or upload it with `curl`.\n\n- For one ChatGPT image attachment, call `create-beauty-content` directly with the\n  file in `image_file`.\n- For multiple attachments, call `upload-file` once per image. Preserve\n  image order, then call `create-beauty-content` once with:\n\n  - the original request in `user_message`;\n  - every returned `fileKey` in `file_keys`;\n  - every returned `fileId` in `file_ids`, aligned by index with `file_keys`.\n\nUse `\"none\"` only when an upload result explicitly returns it as the file ID.\nDo not stop after `upload-file`; it is an intermediate tool.\n\nNever place local paths, upload URLs, file keys, or file IDs inside\n`user_message`. For a large batch, ask the user to prioritize a smaller set or\nprocess it in sequential turns.\n\n### No photo provided\n\nCall `create-beauty-content` with the text-only request so the specialist can provide\na general recommendation. Do not ask the user to upload a photo first. Mention\na photo only as an optional way to receive more personalized visual analysis.\n\nWhen the user chooses to provide a face photo, recommend natural light, a\ncentered face with the hairline and chin visible, and no filters, heavy makeup,\nmasks, or large glasses.\n\n## Continue a session\n\nSave the `sessionId` returned by `create-beauty-content`. Pass it as `session_id` on\nlater YouCam calls in the same workflow. Do not reuse it for unrelated\nrequests, and never invent a session ID.\n\nFor an image already uploaded to or generated by the current session, pass the\nuser's semantic instruction with the existing `session_id`. Include image\nfields only for a new image or when session context is insufficient.\n\nPhoto-based makeup VTO is an exception to semantic session reuse. To apply\nmakeup to a specific photo, the current `create-beauty-content` call needs exactly\none image input: either `image_file` or one matching `file_keys`/`file_ids`\npair. Reuse the previous pair when available. A `session_id` alone is not\nenough for photo-based VTO; without an image input, the App opens live camera\ntry-on.\n\n## Finish the turn\n\nA successful request has two deliverables: the result in the YouCam MCP App\nand a separate assistant text reply in the same turn. This applies even to a\nshort request such as \"幫我換成咖啡廳背景\"; the user does not need to ask for\nan explanation or download link separately. Respect an explicit image-only\nrequest from the user.\n\nInspect the returned status before deciding whether to end the turn:\n\n- For `status=completed`, write the assistant reply after the tool returns,\n  even when the App already displays the image. In the user's language, relay\n  the `answer` and provide a Markdown download link for each `resultFiles`\n  URL, preserving all query parameters. Add a relevant next step from\n  `suggestedNextActions` when available. Do not claim edits absent from the\n  result or invent a download URL. A tool result or widget caption alone does\n  not fulfill the assistant text reply.\n- For `status=processing` with a rendered MCP App, end the current turn and\n  let the App retrieve the pending result. Do not call another result tool,\n  repeat `create-beauty-content`, or describe the pending edit as\n  completed.\n\nIf the App later sends a completion notification for a pending task, use its\nresult data to write the same follow-up reply. Do not start the edit again.\nKeep these response instructions in the assistant workflow; do not append them\nto the image-editing `user_message` sent to YouCam.\n\nRead [MCP errors](references/mcp-errors.md) when authentication, upload,\nprocessing, credit-limit, or upstream calls fail.\n"
}