OpenClaw x Seedream

You have access to RunAPI task tools for Seedream.

Available Seedream models:
- seedream-4.5-edit: 同樣寫實，可透過遮罩／筆刷進行局部區域編輯
  endpoints: /api/v1/seedream/edit_image
  request fields: model, prompt, aspect_ratio, output_quality, source_image_urls
- seedream-4.5-text-to-image: 寫實輸出；人體結構穩定
  endpoints: /api/v1/seedream/text_to_image
  request fields: model, prompt, aspect_ratio, output_quality
- seedream-5-lite-edit: 相較 4.5 更能遵循參考並支援雙語文字排版；速度更快
  endpoints: /api/v1/seedream/edit_image
  request fields: model, prompt, aspect_ratio, output_quality, source_image_urls
- seedream-5-lite-text-to-image: 針對時效性提示詞進行即時網路搜尋
  endpoints: /api/v1/seedream/text_to_image
  request fields: model, prompt, aspect_ratio, output_quality
- seedream-5-pro-edit: 面向精緻成品的高品質寫實圖像編輯
  endpoints: /api/v1/seedream/edit_image
  request fields: model, prompt, aspect_ratio, output_quality, source_image_urls
- seedream-5-pro-layer-decomposition: 將單張圖片拆分為可編輯底圖與透明 PNG 圖層
  endpoints: /api/v1/seedream/decompose_layers
  request fields: model, image_url, size, output_format
- seedream-5-pro-text-to-image: 面向精緻成品的高品質寫實圖像生成
  endpoints: /api/v1/seedream/text_to_image
  request fields: model, prompt, aspect_ratio, output_quality
- seedream-v4-edit: 多圖片編輯，最多 10 張參考圖並支援 4K 輸出
  endpoints: /api/v1/seedream/edit_image
  request fields: model, prompt, aspect_ratio, output_quality, source_image_urls
- seedream-v4-text-to-image: 寫實的 1K-4K 批次生成；每個提示詞最多 6 張圖片
  endpoints: /api/v1/seedream/text_to_image
  request fields: model, prompt, aspect_ratio, output_quality

Use the model ID and endpoint that match the user's request.
Example prompt: 為咖啡店特賣製作一張社群媒體圖片，以粗體呈現「50% OFF」文字，暖色調，4K。
/api/v1/seedream/edit_image: Submit the task, poll for its status, then verify the completed output. POST /api/v1/seedream/edit_image，並儲存回傳的 Task ID。
GET /api/v1/seedream/edit_image/<task-id>，直到狀態為 completed 或 failed。
確認 completed 回應包含 API 參考中記載的端點專屬輸出。
/api/v1/seedream/text_to_image: Submit the task, poll for its status, then verify the completed output. POST /api/v1/seedream/text_to_image，並儲存回傳的 Task ID。
GET /api/v1/seedream/text_to_image/<task-id>，直到狀態為 completed 或 failed。
確認 completed 回應包含 API 參考中記載的端點專屬輸出。
/api/v1/seedream/decompose_layers: Submit the task, poll for its status, then verify the completed output. POST /api/v1/seedream/decompose_layers，並儲存回傳的 Task ID。
GET /api/v1/seedream/decompose_layers/<task-id>，直到狀態為 completed 或 failed。
確認 completed 回應包含 API 參考中記載的端點專屬輸出。