← Back to Skills Marketplace
170
Downloads
1
Stars
0
Active Installs
6
Versions
Install in OpenClaw
/install seedance-waoo-video-pipeline
Description
自动化短视频工作流(story-to-video pipeline):从剧本/分镜到生成、字幕 ASR、TTS、合并交付,支持 Seedance / Vidu / MiniMax 多厂商路由。
Usage Guidance
This skill appears to do what it says, but before installing:
- Review and supply pipeline config (config/pipeline.config.json). The skill expects vendor 'API Key' fields in that config; treat them as secrets.
- Only put API keys for vendors you trust into the config; don't paste unrelated credentials.
- The runtime will read/write files in the workspace and may call out to vendor endpoints to submit/poll/download media. If you want to limit exposure, run it in an isolated/test environment and use test/demo keys first.
- The scripts spawn local binaries (ffmpeg/ffprobe/tsx/node). If you do not want local media processing, ensure those binaries are absent or disabled.
- If you need stronger assurance, inspect any omitted files (remaining 47 files) for network endpoints or unusual code paths; consider running under restricted network or containerized environment.
- Autonomous invocation is allowed by default but the skill's SKILL.md defines explicit checkpoints — still review triggers and only enable runtime autonomy if you are comfortable with that behavior.
Capability Analysis
Type: OpenClaw Skill
Name: seedance-waoo-video-pipeline
Version: 1.1.4
The bundle is a highly structured and professional automation toolkit for a 'story-to-video' pipeline, integrating multiple AI providers like Seedance, Vidu, and MiniMax. It manages a complex workflow including script generation, storyboard planning, video/audio asset production, and final assembly using FFmpeg. The scripts (e.g., `run-seedance-workflow.cjs`, `submit-official-video.cjs`) are well-documented and use secure patterns for process execution, such as passing arguments as arrays to `execFileSync` and `spawnSync` to prevent shell injection. API keys and sensitive configurations are handled appropriately, being sent only to the user-defined service endpoints. No evidence of data exfiltration, malicious prompt injection, or unauthorized system access was found.
Capability Assessment
Purpose & Capability
Name/description match the actual code and instructions: many scripts implement storyboard/prompt generation, first-image, video submit/poll/download, ASR/TTS checks and config guidance. Requiring vendor API keys in the pipeline config (not as environment variables) is expected for a multi-vendor media pipeline.
Instruction Scope
SKILL.md and scripts instruct the agent to read pipeline config/state files, generate artifacts, run local node scripts, spawn helper binaries (ffmpeg/tsx) and call vendor APIs to submit/poll/download media — all within the stated pipeline scope. Checkpoints and user-approval steps are explicit and limit autonomous actions.
Install Mechanism
No install spec; code is instruction-and-script-only. That keeps risk lower because nothing is fetched/installed by an installer step. The code may still download media from configured vendor URLs at runtime (expected behavior).
Credentials
Registry metadata lists no required env vars, and the skill does not demand platform secrets, but it does expect APIKey values inside a JSON pipeline config (config/pipeline.config.json or user-provided path). This is reasonable but note the difference: secrets are stored in config files rather than declared env vars; the skill checks and uses those API keys to call vendor endpoints.
Persistence & Privilege
always is false and the skill is user-invocable. It runs local scripts that read/write files under workspace and may spawn child processes (ffmpeg/node); this is expected for a media pipeline and does not request system-wide persistent privileges.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install seedance-waoo-video-pipeline - After installation, invoke the skill by name or use
/seedance-waoo-video-pipeline - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.1.4
**Prompt engineering overlay and testing scripts introduced:**
- Added `scripts/build-seedance-prompt-pack.cjs` for generating prompt packs with structured blocks and quality scoring.
- Added `scripts/self-test-prompt-pack.cjs` for prompt pack self-testing.
- Updated workflow documentation: prompt engineering overlay is now built-in; now includes prompt quality scoring, scenario fallback, and usage of the new scripts.
- New runtime options for prompt mode, style, camera, scenario, auto-fallback, and minimum score.
- Workflow result outputs now include `promptQuality`, prompt fallback reports, and structured prompt-pack results.
v1.1.3
seedance-waoo-video-pipeline v1.1.3
- Added ASR (Automatic Speech Recognition) configuration check script: `scripts/check-asr-config.ts`.
- Updated documentation to clarify ASR routing preferences and fallback strategies.
- No breaking changes to workflow or interfaces.
v1.1.2
新增能力路由:支持 Seedance / Vidu / MiniMax 多配置映射与自动选路。
新增厂商覆盖参数:--video-vendor、--image-vendor。
优化图文流程:有参考图时默认“原图直绑 first_frame”,减少不必要首图阻塞。
新增首图策略提醒:四件套确认后明确提示可切换 A 原图直绑(默认) / B 图生图。
新增首图策略切换:支持 --first-image-strategy direct|img2img,并可从“换风格/图生图”等指令自动识别。
优化配置引导与就绪检查:start/continue 统一输出配置 guidance,降低误配率。
同步更新技能 Summary(中文),更清晰说明多配置能力。
发布包清理:移除测试配置与本地敏感数据后重打包。
v1.1.1
No file or behavioral changes in this version.
- Version number updated to 1.1.1
- No updates to code or documentation detected
- Functionality and pipeline usage remain the same
v1.1.0
1.新增
支持按能力自动路由(文字生视频 / 图文生视频 / 图生图)。
支持多厂商映射配置:
downstream.waoo.videos.*
downstream.waoo.images.*
支持手动强制指定厂商:
--video-vendor
--image-vendor
支持 runtime.routing.* 路由优先级策略。
2.优化
start/continue 统一接入配置就绪检查与引导输出。
图+文场景默认可直绑 first_frame,减少不必要的首图阻塞。
对“厂商-接口地址”做一致性防呆校验(拦截明显混配,保留自定义网关可用性)。
文档引导升级:配置模板、首跑指南、技能说明同步更新。
3.修复
修复路由写死导致的可扩展性不足问题。
修复部分场景对 image/tts 的过度硬依赖导致流程无法启动的问题。
继续保留并强化 Seedance 官方约束:同请求不混用 first_frame/last_frame 与 reference_image。
v1.0.0
Initial release of seedance-waoo-video-pipeline.
- Added end-to-end story/novel to short-video workflow.
- Supports three modes: minimax_full, vidu_simple, seedance_simple.
- Added provider routing for MiniMax / Vidu / Seedance submit-poll-download chain.
- Added onboarding and first-run guidance.
- Sanitized package for marketplace publishing.
Metadata
Frequently Asked Questions
What is Seedance + Waoo 短视频流水线?
自动化短视频工作流(story-to-video pipeline):从剧本/分镜到生成、字幕 ASR、TTS、合并交付,支持 Seedance / Vidu / MiniMax 多厂商路由。 It is an AI Agent Skill for Claude Code / OpenClaw, with 170 downloads so far.
How do I install Seedance + Waoo 短视频流水线?
Run "/install seedance-waoo-video-pipeline" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Seedance + Waoo 短视频流水线 free?
Yes, Seedance + Waoo 短视频流水线 is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Seedance + Waoo 短视频流水线 support?
Seedance + Waoo 短视频流水线 is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Seedance + Waoo 短视频流水线?
It is built and maintained by wusyu (@wusyu); the current version is v1.1.4.
More Skills