Skill flagged — suspicious patterns detected

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

PPT to Speech Skill

v1.0.3

将 PPT/PPTX 文件转换为结构化演讲稿。当用户说"帮我整理这份PPT"、"把这个PPT转成演讲稿/文章"、"提取PPT内容"、"生成演讲稿"、"PPT转markdown"、"分析这份幻灯片",或提供了 .pptx 文件路径并要求处理时,立即使用此 skill。无需用户配置任何 API Key,由 Agent...

1· 114·0 current·0 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 qigaiai/ppt-to-speech.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "PPT to Speech Skill" (qigaiai/ppt-to-speech) from ClawHub.
Skill page: https://clawhub.ai/qigaiai/ppt-to-speech
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-to-speech

ClawHub CLI

Package manager switcher

npx clawhub@latest install ppt-to-speech
Security Scan
VirusTotalVirusTotal
Suspicious
View report →
OpenClawOpenClaw
Benign
medium confidence
Purpose & Capability
The SKILL.md, the example files, and scripts/extractor_ppt.py all align: the script extracts slide text/notes, converts slides to images, and the agent uses those artifacts to write a speech. The use of LibreOffice + poppler + python-pptx/pdf2image is appropriate for converting PPTX to PDF/images and extracting content.
Instruction Scope
Runtime instructions are scoped to the provided PPTX and generated files (read the generated .md and thumbnails, then write the final .md and package outputs). However, the instructions and script also perform network downloads of runtime dependencies and may invoke system package managers (apt/yum/dnf/pacman) and msiexec; that expands the scope beyond pure local file parsing and requires explicit user awareness/consent.
!
Install Mechanism
There is no install spec, but the script automatically downloads and extracts large binary packages: LibreOffice from an Aliyun mirror (~300MB, tar/zip/msi/dmg) and poppler prebuilt packages (GitHub release zip) and extracts them to user cache/bin directories. The script also may run apt/yum to install poppler-utils. Automatic download+extract of binaries is higher-risk than purely Python-only installs and should be reviewed; the URLs used are recognizable (Aliyun mirror, GitHub releases) but the behavior writes and extracts executables on disk.
Credentials
The skill requests no environment variables or credentials. The only sensitive actions are filesystem and package operations (writing to ~/.cache and the skill's bin dir, modifying PATH temporarily). No unrelated API keys or secrets are requested.
Persistence & Privilege
The skill does not request persistent platform privileges (always:false). It creates files under ~/.cache/ppt_skill and a bin directory under the skill path and may modify PATH for the process. On Linux it may attempt sudo to install poppler-utils via the system package manager — this requires elevated privileges and should be allowed only if you trust the code/environment.
Assessment
This skill appears to do what it claims, but it will: (1) auto-download and extract large binaries (LibreOffice, poppler) from external URLs; (2) attempt to install system packages (poppler-utils) and may require sudo on Linux; (3) write extracted binaries to ~/.cache and the skill bin dir and temporarily modify PATH; (4) require you to run pip install for Python libs. Before installing or running: review the extractor_ppt.py code yourself, or run it in a throwaway VM/container; if you prefer not to allow automatic downloads, manually install LibreOffice and poppler and ensure soffice/pdfinfo are on PATH; verify the download URLs and network policy; and be aware the skill will package and send the generated artifacts to the user (it requires sending the full zip). If you are unsure, test with non-sensitive PPTX files first.

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

latestvk978s69w2t612ak7ttjpe2y88n83m4aa
114downloads
1stars
4versions
Updated 1mo ago
v1.0.3
MIT-0

将 PPTX 文件的每页标题、正文、演讲者备注提取为结构化数据,生成每页缩略图,最终由你(Agent)结合文字内容与缩略图,撰写一篇连贯、有感染力的 Markdown 演讲稿。

输入:用户提供的 PPTX 文件路径 —— $ARGUMENTS


第一步:安装 Python 依赖

首次使用前执行(已安装可跳过):

pip install python-pptx pdf2image Pillow requests

LibreOffice 和 Poppler 由脚本自动检测并下载,无需手动安装。

依赖说明

本技能需要 LibreOfficepoppler-utils 来解析 PPT 并生成缩略图。

首次运行时,脚本会自动:

  • 从阿里云镜像下载 LibreOffice(约 300 MB)并解压到 ~/.cache/ppt_skill/
  • 从 GitHub 下载 poppler 预编译包(约 50 MB)并解压到 ~/.claude/skills/ppt-to-speech/bin/
  • 将上述路径临时添加到当前进程的 PATH

网络访问:仅用于下载上述依赖,所有下载源均为可信镜像。

权限要求

  • Windows:无需管理员权限(使用便携式提取)
  • Linux:若系统中未安装 poppler-utils,脚本会尝试通过 apt/yum 等包管理器安装,可能需要 sudo 权限;若无法获取 sudo,会提示手动安装。

如您不希望自动下载,可提前手动安装 LibreOffice 和 poppler-utils,并确保 sofficepdfinfo 命令在 PATH 中。

第二步:提取内容与生成缩略图

运行脚本,提取 PPT 结构化内容并生成每页缩略图:

python scripts/extractor_ppt.py "$ARGUMENTS"

脚本完成后,输出文件位于与输入文件同目录:

  • <文件名>.md —— 结构化文本(标题 / 正文 / 备注的原始拼接)
  • <文件名>_thumbnails/page_001.png … —— 每页缩略图

第三步:由 Agent 撰写演讲稿

脚本只做机械提取,真正的演讲稿由你来写

3.1 读取提取结果

用 Read 工具读取上一步生成的 .md 文件,获取每页的标题、正文和演讲者备注。

3.2 查看缩略图

用 Read 工具依次查看 _thumbnails/ 目录中的 PNG 图片,理解每页的视觉布局、图表和设计重点,补充纯文字无法传达的信息。

3.3 撰写演讲稿

基于以上内容,按如下要求撰写演讲稿:

结构要求

  • 全文分章节,章节标题对应 PPT 的逻辑段落(不必与页码一一对应)
  • 每个章节开头插入对应页缩略图:![第N页](相对路径/page_00N.png)
  • 以引言开场、以行动号召收尾

写作要求

  • 语言风格:使用口语化但精炼的演讲语言,避免照抄 bullet points。把要点展开成完整句子,必要时用比喻、类比、设问增强感染力。
  • 数据与案例:PPT 中的数据、案例、引述必须完整保留,并用口语化方式解释其意义,这是演讲的血肉。
  • 备注融入:如有演讲者备注(通常在 .md 文件中有单独标记),将其自然融入正文,作为强调或补充说明。
  • 避免逐页罗列:严禁出现“第1页:……;第2页:……”这样的流水账。而是提炼出核心信息,按逻辑重组。

图文要求

  • 结合缩略图中观察到的视觉信息(图表趋势、配色重点、布局层级)进行解读
  • 不要只复述文字,要说明"这张图想说明什么"
  • 在文字描述前或后插入对应的缩略图引用,使读者图文并茂

3.4 写入输出文件

用 Write 工具将最终演讲稿覆盖写入第二步生成的 .md 文件,完成替换。


输出规范

所有产物统一放在一个文件夹内:

产物路径说明
<文件名>.md与输入文件同目录最终 Markdown 演讲稿(由 Agent 撰写)
<文件名>_thumbnails/与输入文件同目录每页 PNG 缩略图(800px 宽,供 Agent 阅读)
输出文件夹名称:<输入文件名(不含扩展名)>_output,位于输入文件同目录或用户指定位置。

处理完成后,必须:

  1. <文件名>.md<文件名>_thumbnails/打包发给用户
  2. 将整个压缩包发送给用户
  3. 不能只发送单个 .md 文件

注意事项

  • 仅支持 .pptx 格式,不支持旧版 .ppt
  • 第三步完全由 Agent 自身完成,无需配置任何外部 API Key
  • 首次运行第二步时会自动下载 LibreOffice(体积较大),请保持网络畅通
  • 缩略图路径在 .md 中使用相对路径,确保文件移动后图片仍可显示

示例文件(./examples/sample_speech.md)

Comments

Loading comments...