Xiao Hong Shu Video Analyser

下载并分析小红书视频内容。当用户提供小红书链接（xiaohongshu.com）时，自动下载视频、提取语音文字、整理总结内容。Use when user provides a xiaohongshu.com URL and wants video content analysis.

Security Scan

The skill largely does what it says (download Xiaohongshu video, extract audio, send to Poe/Gemini for transcription, and summarize), but its metadata omits the required POE_API_KEY and there are a few operational details you should verify before installing.

Purpose & Capability

Name/description match the code and instructions: scripts download a Xiaohongshu video, extract audio (ffmpeg/yt-dlp), and send audio to the Poe API (Gemini) for transcription and summarization. There are no unrelated cloud credentials or unexpected services requested.

Instruction Scope

SKILL.md and scripts describe and implement the expected workflow. The shell script and Python downloader operate only on the provided URL/work directory. The agent is instructed to send audio files to api.poe.com for transcription, which matches the stated design. The scripts read ~/.openclaw/openclaw.json as a fallback for POE_API_KEY.

Install Mechanism

No install spec is present (instruction-only with included scripts). No remote download/install of arbitrary code occurs; only local scripts are run. This is low-risk from an install mechanism standpoint.

Credentials

Metadata lists no required env vars, but both SKILL.md and scripts require a POE_API_KEY (or an entry in ~/.openclaw/openclaw.json). This mismatch is a meaningful omission: the skill will attempt to read the API key from your home config file or environment and will send audio data to Poe. The number and type of credentials requested (a single Poe API key) are proportionate to the task, but the omission from metadata is an incoherence that should be fixed/confirmed before use.

Persistence & Privilege

The skill does not request always:true and does not modify other skills or system-wide settings. It runs transiently and writes outputs to a working directory (default /tmp/xhs-analysis-<ts>), which is expected behavior.

What to consider before installing

This skill appears to implement the stated Xiaohongshu video → audio → Poe/Gemini transcription → summary flow, but double-check these before installing: (1) The published metadata omitted POE_API_KEY even though SKILL.md and the scripts require it — confirm you are comfortable the skill will read $POE_API_KEY or ~/.openclaw/openclaw.json. (2) The skill will upload audio to api.poe.com (a third‑party transcription service) — avoid using it with sensitive or private audio unless you trust the service and key. (3) It depends on ffmpeg (and optionally yt-dlp) being available; yt-dlp is used as a fallback. (4) The downloader scrapes the Xiaohongshu page and writes video/audio/transcripts into a work directory (default /tmp); check and clean that directory if needed. If you plan to use this skill, request the author to update the registry metadata to declare POE_API_KEY as a required environment variable and to document any other runtime requirements. If you don’t trust the source, consider inspecting or running the scripts in an isolated environment or using a throwaway/limited Poe API key.

小红书视频分析器

下载小红书视频并提取语音内容进行总结。

快速使用

当用户提供小红书链接时，执行以下流程：

~/.openclaw/skills/xhs-video-analyzer/scripts/full-analyze.sh "https://www.xiaohongshu.com/explore/..."

工作流程

1. 下载视频 - 使用 Python 脚本或 yt-dlp
2. 提取音频 - 使用 ffmpeg
3. 语音转文字 - 使用 Poe API + Gemini (云端转录)
4. 详细总结 - 基于转录文本生成深度分析报告

依赖

• ffmpeg - 提取音频
• curl - API 请求
• Poe API Key - 需要配置 POE_API_KEY 环境变量或在 ~/.openclaw/openclaw.json 中配置

API 配置

在 ~/.openclaw/openclaw.json 中配置 Poe API：

{
  "models": {
    "providers": {
      "poe": {
        "baseUrl": "https://api.poe.com/v1",
        "apiKey": "YOUR_POE_API_KEY"
      }
    }
  }
}

或设置环境变量：

export POE_API_KEY="your-api-key"

转录技术

使用 Poe API 的 file 格式发送音频到 Gemini 模型进行转录：

{
  "type": "file",
  "file": {
    "filename": "audio.mp3",
    "file_data": "data:audio/mp3;base64,<base64>"
  }
}

输出

• 视频文件: video.mp4
• 音频文件: audio.mp3
• 转录文本: audio.txt
• 音频片段: chunks/ (3分钟一个片段)

详细总结指南

转录完成后，必须对转录文本进行深度分析和详细总结。总结应包含以下内容：

总结结构

1. 视频主题 - 一句话概括视频的核心话题
2. 核心观点 - 提取视频中的主要论点和见解（3-5条）
3. 详细内容分解 - 按话题/章节拆解视频内容，每个部分包含：
   • 子话题标题
   • 具体内容和论据
   • 举例或案例说明
   • 讲述者的个人见解或建议
4. 实用价值 - 视频对观众的实际帮助或启发
5. 关键金句 - 摘录视频中有价值的原话（2-3句）

总结原则

• 详尽而非简略 - 不要只列出要点，要展开说明每个观点的具体内容
• 保留原意 - 准确传达讲述者的观点，不要过度概括导致信息丢失
• 结构清晰 - 用标题、列表、引用等方式组织内容，便于阅读
• 突出价值 - 强调视频中对观众有实际帮助的信息和建议
• 语言自然 - 使用流畅的中文表达，避免机械式罗列

示例格式

```

视频主题

[一句话概括]

核心观点

1. [观点1] - [具体解释]
2. [观点2] - [具体解释]
3. [观点3] - [具体解释]

详细内容

[子话题1]

[详细说明内容，包括背景、论据、案例等]

[子话题2]

[详细说明内容，包括背景、论据、案例等]

[子话题3]

[详细说明内容，包括背景、论据、案例等]

实用价值

• [对观众的实际帮助1]
• [对观众的实际帮助2]

关键金句

"[原话摘录1]" "[原话摘录2]" ```

分析完成后，先输出转录文本的详细总结，再提及工作目录信息。