AGENT SKILL

vidu-s-api

Integrate Vidu S2-Avatar real-time digital characters and Vidu S2-Editing live video editing. Use for Vidu S session creation, Aliyun RTC media, control WebSocket initialization, text instructions, runtime reference-image changes, voice selection, and session cleanup.

License 许可
MIT
最近更新
2026-11-13
Stars
★ 0
Forks
0

01 前言 Front Matter

name
vidu-s-api
description
Integrate Vidu S2-Avatar real-time digital characters and Vidu S2-Editing live video editing. Use for Vidu S session creation, Aliyun RTC media, control WebSocket initialization, text instructions, runtime reference-image changes, voice selection, and session cleanup.

02 资源文件 4

03 SKILL.md 正文

Vidu S API — S2

Upstream skill: shengshu-ai/vidu-s-api. Use the official regional documentation linked below for current product capabilities, limits, and pricing.

Use the user's selected region: China uses api.vidu.cn, Global uses api.vidu.com. Keys are region-specific. Ask for the region only when it is unknown and a live call or region-specific configuration depends on it.

Official documentation: China · Global. Use these for current product capabilities, limits, and pricing. The repository's Node quickstart implements Avatar Real-time and Editing Real-time. For Avatar Component (bring your own ASR/LLM/TTS/RTC) or Offline generation, read the corresponding official docs instead of applying the real-time session flow below.

Select the flow

  • Avatar: create with POST /live/s_avatar/realtime, a character image, persona, and call_mode. Read HTTP reference for payloads and voices when selecting or cloning a voice.
  • Editing: create with POST /live/s_editing/realtime, a reference image and editing_type. Read HTTP reference for payloads and the rendered stream's identity.

Both use server-side Authorization: Token vda_xxx for HTTP and upstream WS. Keep the API key on a trusted server; proxy browser control messages. The optional Editing client_secret returned by creation can remain on that server too.

Treat live IDs as strings. Retain the returned live, rtc, and (Editing only) render_uid. Read the returned duration and token expiry for each session.

Start and maintain a session

  1. Create the chosen session over HTTP.
  2. Open /live/ws/live/connect?live_id=...&conn_id=... through the server, including Editing client_secret when returned. Send conn_init immediately. Read WebSocket protocol before implementing the handshake, retries, or runtime signals.
  3. In Avatar video and Editing, wait for conn_init_ack.success=true, then join Aliyun RTC. Avatar audio can join while control initialization is pending.
  4. For Aliyun Web SDK 7.1.9, call joinChannel(rtc.token, rtc.user_id), using the returned base64 credential directly. Configure the communication channel profile and subscriptions before joining. Vidu supplies credentials; a browser AppKey or self-generated token is unnecessary. Web SDK 7.1.9 has no setDefaultPublishLocalAudioStream or setDefaultPublishLocalVideoStream; use its publishLocal*Stream methods after joining, and setAudioOnlyMode(true) for Avatar audio mode.
  5. Publish microphone audio; publish camera video for Avatar video and Editing. Bind remote video when subscription state is 3 (subscribed): videoSubscribeStateChanged corresponds to camera stream type 1, and screenShareSubscribeStateChanged to screen stream type 2. For Editing, display only render_uid's camera stream. For a single Avatar result view, select the Vidu publisher with a live-bot-* or live-video-push-* user ID and pass the event's stream type to setRemoteViewConfig; unrelated RTC participants must not replace the generated output.
  6. Keep the upstream WS heartbeat alive. The Node ws client replies to ping frames automatically. authInfoWillExpire is an advance warning: log or surface it while keeping the current call active. Release media on authInfoExpired, the actual rtc.token_expire_at deadline, type: 6, or WS close/error. Guard callbacks against events from ended sessions.
  7. On hangup, send type: 5, close WS, stop preview, leave RTC, and destroy the engine. A newly created session must use fresh credentials.

The Node demo exposes Avatar text instructions and Editing reference/scenario changes. The documented Avatar reference-image operation is separate from Editing switch_prompt; use the correct signal when a user requests that capability.

Source differences

The official overview and supplied S2 quickstarts use /live/s_avatar/realtime and /live/s_editing/realtime. Some detail pages still show the older generic creation paths and different Editing authentication examples. Use the named S2 routes and header authentication used by the supplied S2 proxy; do not switch endpoints or expose the API key in browser URLs to follow a conflicting snippet. For Editing NOT_READY, current detail docs specify an automatic success ack; Avatar retries conn_init on the existing socket after 2–3 seconds. Bound startup waiting (the demo uses 60 seconds for Avatar and 120 seconds for Editing).

For an online trial, use https://www.vidu.cn/vidu-stream or https://www.vidu.com/vidu-stream for the user's region.

04 同仓库更多 Skills

endman100
endman100/skill-router-nexus

blender-web-3d

Work with Blender from Claude — headless CLI or the Blender MCP — to build, modify, render, and publish any 3D model as a stunning, fast Three.js/Next.js web experience (gltf-transform meshopt/KTX2, AgX colour pipeline, custom GLSL shaders and material patches, procedural sky/planet, bloom, adaptive quality, runtime batching, baked lightmaps, Vercel deploy). Use whenever the user wants to create, improve, fix, render, or iterate on a Blender model or scene — "improve the model", "make it look stunning", "render this", "make a 3D web demo", "export to GLB", "use blender", "looks wrong in the browser", "optimize the 3D scene", "add shaders", "it's slow", or screenshot-driven feedback on a 3D asset. Also for debugging a Three.js viewer that renders black, glows wrong, or loses/detaches parts, and for choosing Blender MCP vs headless CLI. Ships a runnable demo model and viewer template.

★ 0
endman100
endman100/skill-router-nexus

guizang-ppt-skill

生成横向翻页网页 PPT(单 HTML 文件),含 WebGL 背景、章节幕封、数据大字报、图片网格等模板。提供两种风格:① "电子杂志 × 电子墨水"(衬线 + 流体背景 + 暖色) ② "瑞士国际主义"(无衬线 + 网格点阵 + IKB/柠檬黄/柠檬绿/安全橙高亮)。当用户需要制作分享 / 演讲 / 发布会风格的网页 PPT,或提到"杂志风 PPT"、"瑞士风 PPT"、"Swiss Style"、"horizontal swipe deck"时使用。

图像与多媒体Skill与智能体开发
★ 0
endman100
endman100/skill-router-nexus

youtube-clipper

YouTube 视频智能剪辑工具。下载视频和字幕,AI 分析生成精细章节(几分钟级别), 用户选择片段后自动剪辑、翻译字幕为中英双语、烧录字幕到视频,并生成总结文案。 使用场景:当用户需要剪辑 YouTube 视频、生成短视频片段、制作双语字幕版本时。 关键词:视频剪辑、YouTube、字幕翻译、双语字幕、视频下载、clip video

★ 0
endman100
endman100/skill-router-nexus

slack-gif-creator

Toolkit for creating animated GIFs optimized for Slack, with validators for size constraints and composable animation primitives. This skill applies when users request animated GIFs or emoji animations for Slack from descriptions like "make me a GIF for Slack of X doing Y".

邮件与沟通图像与多媒体
★ 0
endman100
endman100/skill-router-nexus

create-master

基于佛教经典文献,生成特定高僧大德的 AI 教学角色

教师与教育
★ 0
endman100
endman100/skill-router-nexus

writing-dna-skill

从至少 20 篇完整文章中蒸馏可复用的写作 DNA,分析语言、文章结构、选题逻辑、素材策略、认知框架和视觉风格,并生成 Writing-DNA.md;按该 DNA 写作时会读取全部蒸馏产物和 5 篇相关原文。用于中英文作者、账号、品牌或出版物的风格分析与一致性写作。 Distill reusable Writing DNA from at least 20 complete articles for Chinese or English authors, publications, brands, and accounts; use for language, structure, topic logic, source strategy, cognitive-frame, and visual-style analysis.

内容创作者写作与文案Skill与智能体开发
★ 0