Video Analysis Workflow 视频案例分析助手

PassAudited by VirusTotal on May 9, 2026.

Overview

Type: OpenClaw Skill Name: video-analysis-workflow Version: 1.0.0 The skill implements a complex video analysis workflow that involves downloading content from external platforms (Douyin, Bilibili), executing local AI models (Whisper), and running PowerShell scripts with 'ExecutionPolicy Bypass' for environment setup (SKILL.md). While the behavior aligns with the stated purpose of video analysis, the skill introduces significant security risks, including potential command injection via unvalidated URLs/file paths passed to shell tools like yt-dlp and FFmpeg, and the capability to access sensitive browser cookies for authentication, even if the instructions mandate user consent.

Findings (0)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

Granting this permission could expose logged-in session data to the agent or downloader, not just the single video being analyzed.

Why it was flagged

The workflow may access local browser cookies, which are session credentials. Although it requires user authorization, the artifacts do not clearly specify which browser/profile/site cookies are read, whether only the target platform is used, or how cookie data is constrained.

Skill content
如果下载器提示需要 cookies,助手会先询问:

> 是否允许读取浏览器 cookies 重试下载?

没有授权时,不会读取浏览器 cookies。
Recommendation

Only authorize cookie access for a specific, trusted download attempt. Prefer manually downloading the video or using a dedicated browser profile with only the needed platform login.

What this means

Running an unreviewed setup script could change the local environment or install packages beyond what the registry metadata declares.

Why it was flagged

The skill documents a first-use setup script that installs FFmpeg, Python, Whisper, openai-whisper, and yt-dlp, but the supplied manifest does not include that script for review. This is aligned with the video-analysis purpose, but its provenance is not verifiable from the artifacts.

Skill content
powershell -ExecutionPolicy Bypass -File <Skill目录>\scripts\setup-video-case-env.ps1
Recommendation

Inspect the setup script before running it, confirm the package sources, and run it only in an environment where installing media-analysis tools is acceptable.

What this means

Saved transcripts, frames, prompts, and metadata could be reused later or shared with teammates if the case library is synced or exposed.

Why it was flagged

The skill intentionally stores transcripts and structured outputs for later AI workflow reuse. This is purpose-aligned, but it creates persistent local/team knowledge artifacts that may contain private video content or source paths.

Skill content
同时保留 Whisper 原始文件:`.json`、`.srt`、`.txt`、`.vtt`、`.tsv`,方便后续 AI 工作流调用和人工复查。
Recommendation

Store the case library in a controlled location, avoid adding confidential videos unless appropriate, and review generated notes before sharing or syncing them.