Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

Ppt Video

v1.4.1

将PPTX/PDF/HTML演示文稿及背景材料自动转换成1280×720口语化播报视频,支持内容风格识别和语速调节。

0· 91·1 current·1 all-time

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for vincentlau2046-sudo/ppt-video.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "Ppt Video" (vincentlau2046-sudo/ppt-video) from ClawHub.
Skill page: https://clawhub.ai/vincentlau2046-sudo/ppt-video
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install ppt-video

ClawHub CLI

Package manager switcher

npx clawhub@latest install ppt-video
Security Scan
VirusTotalVirusTotal
Suspicious
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
Name/description match the included code: the scripts extract text, render screenshots, run TTS and assemble video. Declared runtime dependencies in SKILL.md (libreoffice, pdftoppm/poppler, ffmpeg, ImageMagick/inkscape, edge-tts) align with the conversion tasks. Minor inconsistencies exist in documented output specs (some docs say 1280×720, others 1024×720; frame rate and resolution differ across README/SKILL.md/package.json and generate.js). These are likely documentation drift rather than malicious intent.
!
Instruction Scope
SKILL.md and scripts instruct the agent to recursively scan the provided input directory for PPT/MD/TXT files and then run external commands (libreoffice, pdftoppm, convert/inkscape, edge-tts, ffmpeg). That behavior is necessary for the task but broad: recursive scanning may pick up unexpected files if the input path is not tightly scoped, and generate.js constructs shell commands by concatenating user-controlled paths into execSync strings. While many variables are wrapped in quotes, concatenation of arbitrary file names into shell commands creates a potential command-injection/escaping risk and could lead to unintended command execution or processing of sensitive files if the input directory is not controlled.
Install Mechanism
No install spec in the registry (instruction-only with code files). The skill relies on common system packages and Python/Node packages documented in SKILL.md/README. No remote downloads or obscure install URLs are used in the metadata; installation is manual via apt/pip and npm runtime, which is proportionate to the functionality.
Credentials
The skill does not request environment variables, secrets, or external credentials in registry metadata. Scripts use standard paths (HOME/.openclaw/workspace examples) and allow optional NODE_BIN override in tests. There are no declared or required API keys—TTS is expected to use the local 'edge-tts' tool which may require network access but no secret is requested by the skill.
Persistence & Privilege
Skill flags are default: always=false and agent invocation allowed. The skill does not request persistent platform privileges and does not attempt to modify other skills or system-wide agent configuration. It writes output to configured directories and temporary files within the project/output directories as expected.
What to consider before installing
This skill appears to do what it claims, but proceed with caution: 1) Provide a tightly-scoped input directory containing only files you intend the skill to process (avoid passing root or broad workspace paths). 2) The Node script uses execSync with concatenated command strings; malicious or specially-crafted filenames might cause command injection or break quoting—sanitize filenames before running, or run the skill in a sandboxed environment (container or VM). 3) The docs disagree on resolution and framerate—verify the produced video matches your requirements before using in production. 4) edge-tts likely requires network access; confirm privacy/usage expectations for TTS output. 5) If you need higher assurance, review the remainder of generate.js (the truncated part) for any network calls, hidden endpoints, or remote uploads. If you lack the ability to sandbox, consider running only on non-sensitive test material or asking the author for clarifications/patches that avoid shell concatenation (use execFile or spawn with argument arrays and strict sanitization).
scripts/generate.js:228
Shell command execution detected (child_process).
Patterns worth reviewing
These patterns may indicate risky behavior. Check the VirusTotal and OpenClaw results above for context-aware analysis before installing.

Like a lobster shell, security has layers — review code before you run it.

latestvk970ghdvjd97w5h8rk4zszkgtn851zbg
91downloads
0stars
2versions
Updated 1w ago
v1.4.1
MIT-0

PPT to Video 技能规范 (v1.4)

将演示文稿 + 背景材料 → 播报/汇报视频

版本: v1.4
作者: Vincent Lau
许可证: MIT
ClawHub 分类: video


📋 技能描述

PPT to Video - 专业演示视频生成工具,支持多种输入格式,自动适配内容风格。

核心流程

PPTX/PDF/HTML + 背景材料 (MD/TXT)
         ↓
    截图 → 风格分析 → TTS → 合成
         ↓
    1024×720 播报视频 (H.264 + AAC)

核心特性

  • 🎯 自动风格识别:新闻/技术/政治/轻松,自动选择 TTS 音色
  • 🎤 统一语速控制:默认 +25% 语速,支持 --rate 参数覆盖
  • 📝 标点符号节奏:使用标点符号(而非特殊标记)控制 TTS 停顿
  • 📊 多格式支持:PPTX/PDF/HTML 输入,自动适配
  • 音画对齐验证:合成后自动检查每页音画同步状态
  • 🗣️ 口语化重写:自动将书面语转为口语化汇报风格(先重点后事实)

📁 文件结构(ClawHub 规范)

ppt-video/
├── SKILL.md                      # ✅ 技能规范文档
├── README.md                     # ✅ 使用说明
├── package.json                  # ✅ 依赖配置
├── LICENSE                       # ✅ 许可证
├── .clawhub/
│   └── _meta.json                # ✅ ClawHub 元数据
├── scripts/
│   ├── generate.js               # ✅ 主脚本
│   └── extract_ppt_text.py       # ✅ PPTX 文本提取
└── templates/                    # 图表模板(可选)

🚀 使用示例

基础用法

# 方式 1: 指定输入目录(自动检测文件类型)
node /home/Vincent/.openclaw/workspace/skills/ppt-video/scripts/generate.js \
  --input /path/to/input/dir

# 方式 2: 已匹配讲稿模式(推荐)
node /home/Vincent/.openclaw/workspace/skills/ppt-video/scripts/generate.js \
  --slides presentation.pptx \
  --script script_matched.md \
  --output ./video/

# 方式 3: 背景材料 + 智能匹配
node /home/Vincent/.openclaw/workspace/skills/ppt-video/scripts/generate.js \
  --slides presentation.pptx \
  --material background.md \
  --readme README.md \
  --output ./video/

# 方式 4: 搜索 note 文件夹
node /home/Vincent/.openclaw/workspace/skills/ppt-video/scripts/generate.js \
  --input /path/to/project/ \
  --searchNotes /path/to/project/ \
  --output ./video/

参数说明

参数必需说明默认值
--date日期标识(用于输出文件名)今天
--input条件输入目录(自动扫描所有文件)-
--slides条件演示文件路径(PPTX/PDF/HTML)-
--script条件已匹配讲稿路径(MD)-
--material条件背景材料路径(MD)-
--readmePPT 说明文件路径(MD)-
--searchNotes搜索 note 文件夹-
--output输出目录~/wechat_articles/Video/ppt-<date>/
--rateTTS 语速(如 "+25%"+25%
--keepTemp保留临时项目文件夹删除

输入优先级

  1. --script 指定已匹配讲稿(最高优先级)
  2. --material + --readme 背景材料 + 智能匹配
  3. --slides 降级模式

📁 输入格式

1. 演示文稿(三选一)

格式支持说明
PPTXPowerPoint 演示文稿(推荐)
PDFPDF 文档
HTML⚠️单页 HTML 幻灯片(简化支持)

2. 背景材料 + README.md(推荐:智能匹配)

方式 A:背景材料带页面对齐标记(最高优先级)

## 第 1 页 封面
欢迎观看本次演示...

## 第 2 页 核心摘要
这里是核心摘要内容...

## 第 3 页 大模型动态
这里是技术详情...

方式 B:背景材料 + README.md(智能匹配,推荐)

背景材料 background.md

## 智谱 GLM-5.1 发布
智谱 AI 正式发布 GLM-5.1,官方称是目前全球最强的开源模型...

## Kimi K2.5 登顶榜单
月之暗面 Kimi K2.5 上线不到 24 小时,登顶 LMarena 开源模型首位...

## 阿里云 CodingPlan
阿里云百炼推出 CodingPlan,四大开源模型自由切换...

PPT 说明 README.md

## 第 1 页 封面
关键词:AI 每日洞察、2026-04-10、零壹情报

## 第 2 页 核心摘要
关键词:智谱 GLM-5.1、Kimi K2.5、AGI 突破

## 第 3 页 大模型动态
关键词:智谱 AI、GLM-5.1、SWE-bench、8 小时持续工作

## 第 4 页 中国厂商
关键词:阿里云、CodingPlan、MiniMax、OpenRouter

## 第 5 页 美国厂商
关键词:OpenAI、Anthropic、Google DeepMind、Genie 3

## 第 6 页 趋势洞察
关键词:开源模型、智能体、长程任务

智能匹配逻辑

  1. 读取 README.md 中每页的关键词
  2. 分析背景材料每段的内容
  3. 自动将段落匹配到关键词最相关的 PPT 页面
  4. 输出匹配报告(每页匹配度分数)

优势

  • ✅ 背景材料无需按 PPT 页数分割
  • ✅ 自动语义匹配,音画对齐
  • ✅ 匹配度可视化,便于调试

🎤 TTS 音色配置

技能自动根据内容风格选择音色:

风格音色语速适用场景
新闻zh-CN-XiaoxiaoNeural+25%新闻播报、日报
技术zh-CN-YunxiNeural+25%技术讲解、汇报
政治zh-CN-YunjianNeural+25%时事政治、国际形势
轻松zh-CN-XiaoyiNeural+25%轻松内容、教育

语速配置

  • 默认统一语速:+25%(所有页面)
  • 支持 --rate 参数覆盖:如 --rate "+30%"
  • 删除封面/结尾特殊加速逻辑(用户反馈不需要)

📤 输出格式

视频规格

属性
分辨率1280×720 (标准 720p 16:9)
视频编码H.264 (libx264)
音频编码AAC (128kbps)
帧率10 fps(静态图片)
格式MP4 (movflags +faststart)

输出目录结构

/home/Vincent/.openclaw/workspace/wechat_articles/Video/ppt-2026-04-09/
├── video_2026-04-09.mp4    # 最终视频
├── screenshots/             # 演示截图(PNG)
│   ├── 01-page.png
│   ├── 02-page.png
│   └── ...
├── audio/                   # TTS 音频(MP3)
│   ├── audio_01.mp3
│   ├── audio_02.mp3
│   └── ...
├── clips/                   # 视频片段(MP4)
│   ├── clip_01.mp4
│   ├── clip_02.mp4
│   └── ...
└── VIDEO_COMPLETE.md        # 完成报告

🔧 依赖要求

必需工具

# Node.js (v18+)
node --version

# edge-tts (Python TTS)
pip install edge-tts
edge-tts --version

# ffmpeg (视频合成)
ffmpeg -version

# LibreOffice (PPTX 截图)
libreoffice --version

# poppler-utils (PDF 截图)
pdftoppm -version

# ImageMagick (HTML 截图,可选)
convert -version

安装命令(Ubuntu/Debian)

# 安装 edge-tts
pip install edge-tts

# 安装 LibreOffice
sudo apt-get install libreoffice

# 安装 poppler-utils
sudo apt-get install poppler-utils

# 安装 ffmpeg
sudo apt-get install ffmpeg

# 安装 ImageMagick
sudo apt-get install imagemagick

🎯 使用场景

1. AI 情报日报视频

node generate.js \
  --input /home/Vincent/.openclaw/workspace/wechat_articles/daily/2026-04-09/

2. 国际形势简报

node generate.js \
  --slides /home/Vincent/.openclaw/workspace/wechat_articles/world_intel_brief/ppt_20260409/international_brief.pptx \
  --material /home/Vincent/.openclaw/workspace/wechat_articles/world_intel_brief/ppt_20260409/script.md \
  --output /home/Vincent/.openclaw/workspace/wechat_articles/world_intel_brief/video_20260409/

3. 技术汇报视频

node generate.js \
  --slides tech_presentation.pptx \
  --material tech_background.md \
  --output ./output/

⚠️ 注意事项

1. 内容风格识别

技能根据关键词自动识别内容风格:

  • 政治:政策、政府、国家、外交、国际形势
  • 技术:技术、代码、API、算法、模型、AI
  • 新闻:新闻、报道、今日、快讯、动态

2. TTS 标点符号控制

edge-tts 不支持特殊停顿标记,使用标点符号控制节奏:

  • 逗号 ,:短停顿
  • 句号 :正常停顿
  • 分号 :中等停顿
  • 破折号 ——:强调停顿

3. 语速配置

  • 默认统一语速:+25%(所有页面)
  • 支持 --rate 参数覆盖:如 --rate "+30%"
  • 删除封面/结尾特殊加速(根据用户反馈优化)

4. 页面对齐验证

技能会自动检查:

  • 讲稿分段数 vs PPT 页数
  • 不匹配时输出警告并建议修复

🛠️ 故障排除

问题 1: LibreOffice 截图失败

错误: PPTX 截图失败

解决:

# 检查 LibreOffice
libreoffice --version

# 如果未安装
sudo apt-get install libreoffice

问题 2: edge-tts 失败

错误: edge-tts: command not found

解决:

pip install edge-tts
edge-tts --version

问题 3: 视频合成失败

错误: ffmpeg: command not found

解决:

sudo apt-get install ffmpeg
ffmpeg -version

📝 版本历史

版本日期变更
v1.42026-04-10✅ 讲稿口语化重写(汇报/新闻风格)<br>✅ 句子长度优化(15-25 字/句)<br>✅ 连接词自动添加(先重点后事实)
v1.22026-04-10✅ 新增 README.md 智能匹配功能<br>✅ 三级优先级策略(标记 > 智能匹配 > 机械分割)<br>✅ 输出匹配度报告(每页分数)
v1.12026-04-10✅ 分辨率改为 1280×720(标准 16:9)<br>✅ 统一语速 +25%,删除封面/结尾特殊加速<br>✅ 强制页面对齐标记(## 第 X 页)<br>✅ 增加音画对齐验证报告
v1.02026-04-09初始版本

🔗 相关技能

  • ppt-master: 专业 PPT 生成
  • professional-pptx-maker: 商务 PPT 生成
  • weixin-video-publish: 微信视频号发布

技能路径: /home/Vincent/.openclaw/workspace/skills/ppt-video/
核心脚本: scripts/generate.js

Comments

Loading comments...