Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Bailian-powered unified creative studio — built-in Wan/Qwen image generation + HappyHorse 1.0 (text-to-video / image-to-video / reference-to-video / video-edit) + Wan 2.6/2.7 + 5 digital-human modes (photo speak / video relip / video reface / pose drive / avatar compose) + CosyVoice & Edge-TTS + storyboard long-video pipeline. Use when the user asks for AI image or video generation on Aliyun Bailian, keyframes, ecommerce images, multi-character video with audio sync, video editing/style transfer
.claude/skills/openakita-happyhorse-video/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 38% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 135% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 46% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 79% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 86% | 0% |
> 一线插件。阿里云百炼万相/千问图片 + HappyHorse 1.0 + 万相 2.6/2.7 + s2v + animate + > videoretalk + cosyvoice + qwen-vl 全链路。OSS 签名 URL 喂 DashScope。
代理在用户出现以下意图时优先调用 happyhorse-video 的工具：
不应触发 happyhorse-video：
seedance-videohh_image_*；若用户明确要求通义生图独立应用，再调 tongyi-imagecosyvoice 工具或主端 TTS| 工具名 | 用途 | 关键参数 | |---|---|---| | hh_image_create | 文生图片 | prompt, model_id?, size?, n? | | hh_image_edit | 图像编辑/多图融合 | prompt, images[] 或 from_asset_ids, model_id? | | hh_image_style_repaint | 风格重绘 | images[] 或 from_asset_ids, style_index? | | hh_image_background | 背景生成 | images[] 或 from_asset_ids, prompt/ref_prompt? | | hh_image_outpaint | 画面扩展 | images[] 或 from_asset_ids, output_ratio? | | hh_image_sketch | 涂鸦作画 | prompt, images[] 或 from_asset_ids, sketch_style? | | hh_image_ecommerce | 电商场景图 | product_name, prompt, ecommerce_scenes[] | | hh_t2v | 文生视频 | prompt, model?, duration?, resolution?, aspect_ratio? | | hh_i2v | 图生视频（首帧 / 首尾帧 / 续写） | prompt, first_frame_url 或 from_asset_ids, last_frame_url?, task_type? | | hh_r2v | 参考生视频（多角色互动） | prompt, reference_urls 或 from_asset_ids, shot_type? | | hh_video_edit | 视频编辑 | prompt, source_video_url（兼容 video_url）, reference_urls? | | hh_photo_speak | 照片说话 | image_url, text 或 audio_url, voice_id? | | hh_video_relip | 视频换嘴 | source_video_url（兼容 video_url）, text 或 audio_url, voice_id? | | hh_video_reface | 视频换人 | image_url, source_video_url（兼容 video_url）, mode_pro? | | hh_pose_drive | 图生动作 | image_url, source_video_url（兼容 video_url）, mode_pro? | | hh_avatar_compose | 数字人合成 | image_url + image_urls[]（兼容 ref_images_url[]）, prompt, text/audio_url | | hh_long_video_create | 长视频分镜拼接 | story, total_duration, mode (serial/parallel) | | hh_cost_preview | 估算费用，不实际提交 | 各 mode 必需的素材或 text_chars/audio_duration_sec | | hh_status | 查询任务状态 | task_id | | hh_list | 列最近任务 | limit? |
工具返回 JSON 包含：ok, task_id, status, mode, model_id, video_url, video_path, last_frame_url, last_frame_path, image_urls, local_paths, asset_ids。失败时 ok=false + error_kind + error_message + terminal=true。
每个 hh_* 创建工具的输入 schema 都接受 from_asset_ids: string[]， runtime 把上游图片工作台（如 tongyi-image）产出的 asset_ids 透传过来：
| mode | from_asset_ids 角色映射 | |---|---| | i2v | [0] → first_frame，[1+] → reference_image | | i2v_end | [0] → first_frame，[1+] → last_frame | | r2v | 全数组 → reference_urls（按角色顺序） | | video_edit / video_extend / video_relip | [0] → source_video_url，[1+] → reference_urls | | video_reface | [0] → source_video_url，[1] → image_url | | pose_drive | [0] → image_url，[1] → source_video_url | | avatar_compose | [0] → image_url，[1+] → image_urls | | image_* | 全数组 → images，[0] → image_url |
工具返回的 asset_ids 是本插件登记到 Asset Bus 的图片/视频资产，下游可继续消费。
| 模型族 | 协议 | 关键参数 | |---|---|---| | happyhorse-1.0-* | 新版 video-synthesis | resolution: "720P" / "1080P"（P 大写）。禁用 with_audio / size / quality / fps / audio | | wan2.7-i2v | 新版 video-synthesis + 多模态 | resolution, task_type (first-frame/first-and-last-frame/video-continuation) | | wan2.6-* | 旧版 video-synthesis | size: "1280*720"（星号 WH）, audio: true | | wan2.2-s2v / s2v-detect | s2v 旧版 | OSS 签名 URL 必填 | | wan2.2-animate-* | animate 旧版 | mode: "wan-std"\|"wan-pro" | | videoretalk | videoretalk 旧版 | 视频 + 音频 URL |
text1. Settings 填 DashScope API Key + OSS 四件套 2. Create Tab → t2v → prompt: "海上日出，慢镜头，电影感" 3. duration=5, resolution=720P → 估算费用 ≈ ¥4.5（HappyHorse 1.0） 4. 「直接提交」→ Tasks Tab 看进度 → 60-180 秒后 done 5. 抽屉看视频，下载 mp4
Semaphore(1) 串行化。hh_status 返 not_found；本地 sqlite 仍存 metadata。tongyi-image — 上游分镜图工作台（asset_ids → from_asset_ids）seedance-video — 火山版视频工作台（双轨并存，不替代）avatar-studio — 数字人专项（多 backend，不替代）| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 14,140 | 3,745 | -74% | 1 | 1 | 0% | 2,277 | 2,610 | +15% | 0 | 0 | — |
case-02 | fail→fail | 18,547 | 15,975 | -14% | 1 | 1 | 0% | 2,802 | 3,526 | +26% | 0 | 0 | — |
case-03 | fail→pass | 15,550 | 5,745 | -63% | 1 | 1 | 0% | 2,231 | 3,073 | +38% | 0 | 0 | — |
case-04 | pass→pass | 14,648 | 3,690 | -75% | 1 | 1 | 0% | 2,758 | 2,699 | -2% | 0 | 0 | — |
case-05 | pass→pass | 11,461 | 4,302 | -62% | 1 | 1 | 0% | 1,821 | 2,811 | +54% | 0 | 0 | — |
case-06 | fail→pass | 7,538 | 3,678 | -51% | 1 | 1 | 0% | 1,159 | 2,723 | +135% | 0 | 0 | — |
case-07 | fail→pass | 11,281 | 4,873 | -57% | 1 | 1 | 0% | 2,130 | 3,111 | +46% | 0 | 0 | — |
case-08 | pass→pass | 10,502 | 3,920 | -63% | 1 | 1 | 0% | 1,854 | 2,910 | +57% | 0 | 0 | — |
case-09 | pass→pass | 11,664 | 3,344 | -71% | 1 | 1 | 0% | 2,077 | 2,770 | +33% | 0 | 0 | — |
case-10 | fail→pass | 9,481 | 3,785 | -60% | 1 | 1 | 0% | 1,622 | 2,896 | +79% | 0 | 0 | — |
case-11 | fail→pass | 9,469 | 3,746 | -60% | 1 | 1 | 0% | 1,470 | 2,738 | +86% | 0 | 0 | — |
case-12 | fail→pass | 13,606 | 4,222 | -69% | 1 | 1 | 0% | 2,010 | 2,823 | +40% | 0 | 0 | — |
case-13 | fail→pass | 15,867 | 4,762 | -70% | 1 | 1 | 0% | 2,462 | 2,958 | +20% | 0 | 0 | — |
case-14 | fail→pass | 13,963 | 22,627 | +62% | 1 | 1 | 0% | 2,006 | 2,759 | +38% | 0 | 0 | — |
case-15 | fail→pass | 16,212 | 6,511 | -60% | 1 | 1 | 0% | 2,227 | 3,161 | +42% | 0 | 0 | — |
case-16 | fail→pass | 11,891 | 4,080 | -66% | 1 | 1 | 0% | 1,756 | 2,877 | +64% | 0 | 0 | — |
case-17 | fail→pass | 15,786 | 5,938 | -62% | 1 | 1 | 0% | 2,367 | 3,164 | +34% | 0 | 0 | — |
case-18 | fail→pass | 15,788 | 5,309 | -66% | 1 | 1 | 0% | 2,187 | 3,096 | +42% | 0 | 0 | — |
case-19 | fail→pass | 10,272 | 4,163 | -59% | 1 | 1 | 0% | 1,674 | 2,744 | +64% | 0 | 0 | — |
case-20 | fail→pass | 15,379 | 6,454 | -58% | 1 | 1 | 0% | 2,267 | 3,222 | +42% | 0 | 0 | — |
case-21 | fail→pass | 10,985 | 2,797 | -75% | 1 | 1 | 0% | 1,640 | 2,628 | +60% | 0 | 0 | — |
case-22 | fail→pass | 12,815 | 3,710 | -71% | 1 | 1 | 0% | 1,965 | 2,802 | +43% | 0 | 0 | — |
case-23 | pass→pass | 8,795 | 3,188 | -64% | 1 | 1 | 0% | 1,566 | 2,763 | +76% | 0 | 0 | — |
case-24 | fail→pass | 7,025 | 4,264 | -39% | 1 | 1 | 0% | 1,187 | 2,970 | +150% | 0 | 0 | — |
case-25 | fail→pass | 19,082 | 2,007 | -89% | 1 | 1 | 0% | 2,926 | 2,471 | -16% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 25 cases were attempted. The headline lift of +72 percentage points is the difference between those two pass rates over the 25 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.