analyze video by qwen

使用 Qwen 3.5 Plus 模型分析视频内容，支持本地文件和远程 URL，可自定义分析提示词和抽帧频率

Security Scan

The skill's code and instructions are consistent with its stated purpose (sending videos to the DashScope/Qwen multimodal API for analysis); nothing in the bundle requests unrelated credentials or privileged persistence, but local video files will be transmitted to the external API and you should verify the dashscope package/service and API key before use.

Purpose & Capability

Name/description state 'analyze video with Qwen'; the included script (scripts/analyze.py) and requirements (dashscope) implement that behavior by calling a DashScope MultiModalConversation with a qwen3.5-plus model. The script reads an API key from ~/.openclaw/openclaw.json as documented. Minor metadata mismatches (/_meta.json owner/version vs registry metadata) are present but do not affect capability coherence.

Instruction Scope

SKILL.md and the script only instruct running the bundled Python script against a local path or remote URL and to store the DashScope API key in ~/.openclaw/openclaw.json. Important privacy note: for local files the script constructs a file:// URL and relies on the DashScope client to access/upload the file — local video content will be transmitted to the remote DashScope service for analysis, which the SKILL.md does not explicitly warn about.

Install Mechanism

No install spec is provided (instruction-only with a code file). A requirements.txt lists 'dashscope' so the user must install that Python package; there is no arbitrary download or archive extraction. This is proportionate for a Python-based API client.

Credentials

The skill does not request environment variables or unrelated credentials. It reads a single API key from ~/.openclaw/openclaw.json (skills.dashscope.apiKey), which matches the need to call the DashScope API. No other files or secrets are accessed.

Persistence & Privilege

always is false, the skill is user-invocable and not requesting persistent/always-on privileges. The included .claude/settings.local.json grants permission to run python/bash which aligns with the script-based usage; the skill does not modify other skills or request system-wide configuration changes.

Assessment

This skill appears to do what it says: it calls the DashScope/Qwen multimodal API to analyze videos. Before installing or running: 1) Know that local videos will be uploaded to the external DashScope service — do not use it on sensitive/personal videos unless you trust that service and account. 2) Verify the dashscope Python package (source/maintainer) and confirm you will install it in a controlled environment. 3) Provide an API key in ~/.openclaw/openclaw.json as described (the script will exit if missing). 4) Note small metadata/version mismatches in bundled files (non-security-critical). If you need a privacy-preserving local-only analysis, this skill is not suitable because it uses a remote model/service.

Analyze Video with Qwen

Overview

使用阿里云 Qwen 3.5 Plus 多模态模型对视频进行智能分析。支持本地视频文件和远程 URL，可自定义分析问题和视频抽帧频率（FPS）。

何时使用

• 需要理解视频内容、场景描述
• 需要视频中的动作识别、物体检测
• 需要生成视频摘要或分析报告
• 需要分析远程在线视频

快速用法

分析本地视频

默认设置分析视频：

python scripts/analyze.py /path/to/video.mp4

自定义提示词：

python scripts/analyze.py /path/to/video.mp4 --prompt "请详细描述视频中的每个场景"

自定义抽帧频率（FPS越高，分析越精细）：

python scripts/analyze.py /path/to/video.mp4 --fps 5

分析远程视频 URL

直接分析远程视频：

python scripts/analyze.py https://example.com/video.mp4

组合使用：

python scripts/analyze.py /path/to/video.mp4 --fps 3 --prompt "视频中出现了哪些人物和物体？"
python scripts/analyze.py https://example.com/video.mp4 --fps 4 --prompt "请详细描述视频场景"

参数说明

参数	说明	默认值	必填
video_source	视频文件路径或远程 URL（支持 http/https）	-	是
--fps	抽帧频率，每秒抽取的帧数	2	否
--prompt	分析提示词	"这段视频描绘的是什么景象？"	否

配置

API Key 从 ~/.openclaw/openclaw.json 的 skills.dashscope.apiKey 字段读取。

如未配置，请添加以下内容：

{
  "skills": {
    "dashscope": {
      "apiKey": "your-dashscope-api-key"
    }
  }
}

备注

• 本地视频路径可以是绝对路径或相对路径
• 远程视频 URL 必须是可公开访问的直链
• FPS 越高，API 调用成本越高，建议根据视频长度和需求调整