← Back to Skills Marketplace
ricksf

Anime Drama

by rickSF · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ suspicious
177
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install anime-drama
Description
将小说原文自动转换为动漫短剧。用户输入故事文本,系统自动完成:分镜脚本生成 → 文生图(RH AI应用)→ 图生视频(RH AI应用)→ ffmpeg合并成片。支持竖屏9:16短视频输出。
README (SKILL.md)

动漫短剧生成技能 (Anime Drama Pipeline)

输入小说原文 → 输出竖屏动漫短视频(分镜脚本 + 图片 + 视频片段 + 合并成片)

核心流程

小说原文
  ↓
分镜脚本生成(按段落拆分为多个镜头)
  ↓
并行/逐个执行:
  ├─ 文生图(RH AI应用 AppId: YOUR_IMAGE_APP_ID,节点 12:value)
  └─ 图生视频(RH AI应用 AppId: YOUR_VIDEO_APP_ID,节点 325:value + 269:image)
  ↓
ffmpeg 合并所有视频片段
  ↓
最终成片(竖屏 1080×1920,MP4)

必读参考文档

  • {baseDir}/references/pipeline-usage.md — 完整使用说明
  • {baseDir}/references/ffmpeg-install.md — ffmpeg 安装指南
  • {baseDir}/references/runninghub-ai-app-notes.md — RH AI应用调用要点

核心脚本

python3 {baseDir}/scripts/anime_drama_pipeline.py \x3C小说原文文件> [输出目录]

Pipeline 脚本位置

{baseDir}/scripts/anime_drama_pipeline.py

RH API 关键参数(永久记忆)

用途 AppID / 端点 节点
文生图 YOUR_IMAGE_APP_ID 12:value(单节点,直接传 prompt)
图生视频 YOUR_VIDEO_APP_ID 325:value(文本)+ 269:image(图片)

图生视频调用前必须先上传图片:

  1. POST /task/openapi/upload 上传本地图片
  2. 取返回值中的 fileName(格式 api/xxx.png
  3. fileName 作为 269:image 的值传入

视频合并(ffmpeg)

# 生成文件列表
cat > /tmp/video_list.txt \x3C\x3C 'EOF'
file '/path/shot_001_vid.mp4'
file '/path/shot_002_vid.mp4'
EOF

# 合并(竖屏 1080×1920)
ffmpeg -y -f concat -safe 0 -i /tmp/video_list.txt \
  -c:v libx264 -pix_fmt yuv420p \
  -vf "scale=1080:1920:force_original_aspect_ratio=decrease,pad=1080:1920:(ow-iw)/2:(oh-ih)/2:black" \
  output.mp4

ffmpeg 路径(已安装到): /usr/local/lib/python3.10/dist-packages/imageio_ffmpeg/binaries/ffmpeg-linux-x86_64-v7.0.2

分镜逻辑

默认按 \ 换行分割,每段 = 一个镜头,每个镜头生成一张图 + 一段视频。 如需更智能的分镜,可传入 LLM 优化每个镜头的描述文字。

输出格式

  • 图片:PNG,~1.7-2.0MB/张
  • 视频:MP4(H.264 + AAC),每个镜头默认 5 秒
  • 最终成片:MP4,1080×1920(竖屏),16:9 源视频自动 pad 为竖屏

常见错误处理

问题 原因 解决方法
文生图失败 AI 应用未在网页端运行过 先在 RH 网页端手动运行一次应用
图生视频失败 图片未上传到 RH 必须先调用 /task/openapi/upload 获取 fileName
ffmpeg: command not found 未安装 ffmpeg 使用 imageio-ffmpeg 的内置 ffmpeg
超时中断 视频生成需 2-5 分钟/镜头 脚本可中断后继续运行(已生成的不会重复生成)

中断后继续

脚本支持断点续传:已生成的图片/视频不会重复生成,中断后重新运行即可继续。

Usage Guidance
This skill appears to do what it says, but the supplied Python script constructs shell commands that include text from your story and executes them with shell=True — this creates a command-injection vulnerability if any input is untrusted. Recommendations before installing/using: 1) Review runninghub_app.py (the script it invokes) and ensure it's trusted. 2) Run the pipeline in an isolated, non-privileged environment (container or VM). 3) Do not feed untrusted text into the pipeline; sanitize or escape prompts (use shlex.quote or subprocess.run with argument lists instead of shell=True). 4) Limit/rotate the RUNNINGHUB_API_KEY and consider scoping it to a test account. 5) Verify any manually downloaded ffmpeg binary before use. If you cannot audit or fix the subprocess usage, treat the skill as untrusted and avoid running it on sensitive hosts or with sensitive credentials.
Capability Analysis
Type: OpenClaw Skill Name: anime-drama Version: 1.0.0 The skill provides a pipeline for generating anime videos using the RunningHub API and ffmpeg. It is classified as suspicious due to a critical shell injection vulnerability in `scripts/anime_drama_pipeline.py`, where user-provided story text (prompts) is passed directly into shell commands via `subprocess.run(shell=True)`. While the logic appears to align with the stated purpose and no evidence of intentional malice (such as data exfiltration or backdoors) was found, the lack of input sanitization allows for arbitrary command execution. The skill also depends on an external script (`runninghub_app.py`) and references external binaries in its documentation.
Capability Assessment
Purpose & Capability
Name/description align with required binaries (python3, curl, ffmpeg) and the RUNNINGHUB integration. The script and docs consistently implement story→image→video→ffmpeg pipeline and declare dependency on a 'runninghub' helper.
Instruction Scope
SKILL.md and the script stay within the stated purpose, but the implementation constructs shell command strings that embed user story/prompt text (enhanced_prompt) and then executes them via subprocess.run(..., shell=True). This allows command injection if the story or prompts contain shell metacharacters. The script also assumes another skill's script exists at ~/.openclaw/workspace/skills/runninghub/scripts/runninghub_app.py and will call it with the API key.
Install Mechanism
No automated install spec (instruction-only + shipped script), so nothing gets implicitly downloaded/installed by the platform. References include manual download of a static ffmpeg binary from johnvansickle.com (documented, not automated) — verify any manual binary downloads before running.
Credentials
Only RUNNINGHUB_API_KEY (primaryEnv) and optional RH_IMAGE_APP_ID / RH_VIDEO_APP_ID are used; these are appropriate for the RunningHub API calls. The script posts images to https://www.runninghub.cn which matches the declared integration.
Persistence & Privilege
always:false and no attempt to change other skills/config. The skill writes workspace files under the user's home directory (anime-drama-workspace), which is expected for generated artifacts but means you should run under a non-privileged account or sandbox.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install anime-drama
  3. After installation, invoke the skill by name or use /anime-drama
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Anime-drama v1.0.0 – Initial Release - Automatically converts novel text into vertical-format (9:16) anime-style short videos. - Complete pipeline: storyboard generation → text-to-image → image-to-video → video merge via ffmpeg. - Supports input of any novel/story text; splits into scenes automatically. - Integrates with RunningHub AI for image and video generation. - Output: MP4 video, 1080×1920 resolution, with robust error handling and resumable workflow.
Metadata
Slug anime-drama
Version 1.0.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is Anime Drama?

将小说原文自动转换为动漫短剧。用户输入故事文本,系统自动完成:分镜脚本生成 → 文生图(RH AI应用)→ 图生视频(RH AI应用)→ ffmpeg合并成片。支持竖屏9:16短视频输出。 It is an AI Agent Skill for Claude Code / OpenClaw, with 177 downloads so far.

How do I install Anime Drama?

Run "/install anime-drama" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Anime Drama free?

Yes, Anime Drama is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Anime Drama support?

Anime Drama is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Anime Drama?

It is built and maintained by rickSF (@ricksf); the current version is v1.0.0.

💬 Comments