← 返回 Skills 市场
jackyken

CreateVideo Podcast to Video

作者 Conan · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ suspicious
91
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install create-podcast-video-universal
功能描述
视频生成工具。当用户说"CreateVideo"、"创建视频"、"生成视频"或提供文案要求制作视频时触发。支持文本转语音(通过 ListenHub MCP)、模版视频裁剪合并、内容分析输出。依赖 ffmpeg 和 ListenHub MCP Server。
使用说明 (SKILL.md)

CreateVideo Skill

功能说明

将音频或文本转化为带背景音乐的竖屏视频。

支持两种输入模式:

  • 模式A:用户直接提供音频文件(MP3)
  • 模式B:用户提供文本文案,系统自动调用 TTS 生成音频

工作流程

步骤1:接收输入

用户提供以下任一:

  • 音频文件:MP3/WAV 格式的配音
  • 文本文案:需要转化为语音的文案内容

步骤2:创建任务目录

DATE=$(date +%Y%m%d)
BASE_DIR="{{skills_dir}}/../video-projects"
TASK_DIR="$BASE_DIR/${DATE}_001"

COUNTER=1
while [ -d "$TASK_DIR" ]; do
  COUNTER=$((COUNTER + 1))
  TASK_DIR="$BASE_DIR/${DATE}_$(printf "%03d" $COUNTER)"
done

mkdir -p "$TASK_DIR/temp"

{{skills_dir}} 为 skill 所在目录,安装后会自动替换为实际路径。

步骤3:准备视频模版

将背景视频放入 video-projects/ 目录(建议 720×1280 竖屏,MP4 格式)。

推荐模版来源(免费素材):

  • Pexels / Pixabay 的竖屏视频
  • 自己拍摄的背景视频

步骤4:获取音频(二选一)

模式A:用户提供音频文件

将用户发来的音频文件保存到 $TASK_DIR/voice.mp3

模式B:文本转语音(TTS)

调用 ListenHub MCP 的 create_flowspeech 接口:

工具:create_flowspeech
参数:
  - sourceType: "text"
  - sourceContent: 用户文案
  - speakerId: 用户指定的音色ID(在 ListenHub 音色库中选择)
  - language: "zh"
  - mode: "smart"

首次使用前,用户需在 ListenHub 音色库中选择自己喜好的音色,复制其 ID 后使用。

步骤5:合并视频与音频

DURATION=$(ffprobe -v error -show_entries format=duration -of default=noprint_wrappers=1:nokey=1 "$TASK_DIR/voice.mp3")

ffmpeg -y \
  -stream_loop 3 -i "$TASK_DIR/template.mp4" \
  -i "$TASK_DIR/voice.mp3" \
  -t $DURATION \
  -c:v libx264 -preset fast -crf 23 \
  -c:a aac -map 0:v:0 -map 1:a:0 \
  "$TASK_DIR/output.mp4"

说明:

  • -stream_loop 3 表示循环4次,根据音频长度调整
  • 模版视频时长应足够覆盖音频长度,或使用 math.ceil(DURATION/TEMPLATE_DUR) 计算循环次数
  • 如服务器不支持 libx264 编码,可使用 -c:v copy 直接复制原视频流

步骤6:生成内容分析(仅模式B需要)

基于原始文案生成并保存到 $TASK_DIR/内容分析.txt

  1. 核心金句 — 最具传播力的一句话
  2. 洞见 — 深层洞察和价值点
  3. 内容描述 — 50字概括视频核心内容
  4. 朋友圈文案 — 200字推广文案

步骤7:输出结果

**视频创建完成!**

**任务目录**: {路径}
**最终视频**: output.mp4

**音频来源**: {TTS合成 / 用户音频}
**时长**: {时长}

默认配置

配置项 说明
基础目录 {skills_dir}/../video-projects
模版目录 video-projects/
音频文件 voice.mp3
输出文件 output.mp4

目录结构

video-projects/
├── v_template.mp4      # 背景视频模版(需用户自行准备)
├── 20260411_001/
│   ├── voice.mp3       # 配音音频
│   ├── template.mp4    # 复制的模版
│   └── output.mp4      # 最终视频
└── temp/               # 临时文件(可定期清理)

依赖

  • ffmpeg / ffprobe — 系统安装,视频处理
  • ListenHub MCP Server — TTS 语音合成(如使用模式B)

ListenHub 安装配置(模式B必需)

1. 安装 MCP Server

npx -y @marswave/listenhub-mcp-server

2. 配置到 OpenClaw

在 OpenClaw 配置中添加 MCP Server 连接。

3. 获取音色

调用 get_speakers 工具(language=zh)查看可用音色列表,选择音色后复制其 ID。

注意事项

  1. 模版视频需用户自行准备,建议使用竖屏(9:16)格式
  2. 模版视频放入 video-projects/ 目录即可被自动使用
  3. 每次任务自动创建日期序号目录
  4. 中间文件在 temp/ 子目录,可定期清理
  5. 如服务器不支持视频编码,可使用 -c:v copy 模式(需要模版视频编码格式兼容)
安全使用建议
This skill appears to implement a reasonable podcast→video flow, but several things don't add up: (1) the registry lists no required binaries or installs, yet the instructions require ffmpeg/ffprobe and ask you to install a ListenHub MCP server via npx (which will download/run code). (2) It asks you to add the MCP server to OpenClaw config (likely requiring host/credentials) but does not declare what credentials are needed. Before installing or running: verify and trust the npm package @marswave/listenhub-mcp-server (review its npm page/source), ensure ffmpeg/ffprobe are installed from trusted sources, run the MCP server and any new installs in a sandbox if possible, and confirm what configuration/credentials will be stored in your OpenClaw instance. If you need higher assurance, ask the publisher for an explicit install manifest and a list of required credentials and where they are stored.
功能分析
Type: OpenClaw Skill Name: create-podcast-video-universal Version: 1.0.0 The skill bundle is a legitimate tool for generating videos from text or audio. It uses standard media utilities like ffmpeg and ffprobe for video processing and integrates with the ListenHub MCP server for text-to-speech functionality. The shell commands in SKILL.md are used appropriately for directory management and media encoding, with no evidence of data exfiltration, unauthorized access, or malicious intent.
能力评估
Purpose & Capability
The description says the skill depends on ffmpeg/ffprobe and ListenHub MCP, and the runtime instructions use ffmpeg/ffprobe and call ListenHub TTS. However the registry metadata and requirements list no required binaries, env vars, or install steps. This mismatch (declared-as-none vs instructions-that-require-system tools and an MCP server) is incoherent and should be justified.
Instruction Scope
The SKILL.md instructs the agent to create directories under {skills_dir}/../video-projects, save user audio, run ffprobe/ffmpeg commands, and to install/run a ListenHub MCP server via npx and add it to OpenClaw config. These are within a video-generation scope, but the instructions also tell the agent to download and run external code (npx @marswave/listenhub-mcp-server) and modify OpenClaw configuration — actions with side effects that are not declared in registry metadata.
Install Mechanism
There is no formal install spec, but the docs explicitly advise running 'npx -y @marswave/listenhub-mcp-server', which will fetch and run code from the npm registry. That is a moderate-risk download/install step that is not captured in the skill's install metadata. ffmpeg/ffprobe are also required by the instructions but not declared as required binaries.
Credentials
The skill declares no required environment variables or credentials. However it requires configuring a ListenHub MCP Server connection in OpenClaw (likely involving endpoints/credentials) and references selecting speaker IDs — the required configuration/credentials are not documented in the skill metadata. No explicit secret exfiltration is requested, but missing credential requirements are a gap.
Persistence & Privilege
The skill is not always-enabled, is user-invocable, and does not request system-wide privileges. It writes files under a project directory relative to the skill (video-projects) and suggests adding a service to the agent config, which are expected behaviors for this functionality. Autonomous invocation is allowed by default but not by itself a red flag here.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install create-podcast-video-universal
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /create-podcast-video-universal 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.0
Initial release of create-video skill: - Generates vertical videos from either audio files or text (with TTS synthesis). - Automates project directory creation and manages input assets. - Supports merging audio with background video templates using ffmpeg. - Provides automatic content analysis outputs for TTS cases. - Requires ffmpeg/ffprobe and (for TTS) ListenHub MCP Server. - Organizes outputs in timestamped folders for easy management.
元数据
Slug create-podcast-video-universal
版本 1.0.0
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 1
常见问题

CreateVideo Podcast to Video 是什么?

视频生成工具。当用户说"CreateVideo"、"创建视频"、"生成视频"或提供文案要求制作视频时触发。支持文本转语音(通过 ListenHub MCP)、模版视频裁剪合并、内容分析输出。依赖 ffmpeg 和 ListenHub MCP Server。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 91 次。

如何安装 CreateVideo Podcast to Video?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install create-podcast-video-universal」即可一键安装,无需额外配置。

CreateVideo Podcast to Video 是免费的吗?

是的,CreateVideo Podcast to Video 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

CreateVideo Podcast to Video 支持哪些平台?

CreateVideo Podcast to Video 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 CreateVideo Podcast to Video?

由 Conan(@jackyken)开发并维护,当前版本 v1.0.0。

💬 留言讨论