Video Frame Analyzer
PassAudited by ClawScan on May 12, 2026.
Overview
This skill appears to do what it claims—extract and analyze frames from user-provided videos—but it runs local Python helpers and may send video frames to a multimodal model.
Before installing, be aware that the skill runs local Python scripts, may require installing video-processing packages, writes extracted frames and reports to disk, and sends selected frames to the multimodal model you choose. It appears purpose-aligned and benign, but only analyze videos you are allowed and comfortable to process.
Findings (3)
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.
The helper can read the specified video and write extracted frame images into the chosen output directory.
The skill instructs the agent to run a local Python helper. This is central to the frame-extraction purpose and the included scripts do not show network, credential, or destructive behavior, but it still executes code with the user's local permissions.
用 Bash 工具执行帧提取脚本: python "$SKILL_DIR/smart_extract.py" "你的视频路径.mp4" "视频名_frames/" 15
Run it only on intended video files, confirm output paths, and avoid granting elevated shell privileges.
A path collision or unexpected local file could cause the wrong helper script to be run, and unpinned package installs may change over time.
The helper path is resolved by taking the first matching smart_extract.py under the skills directory, and dependencies are documented as pip installs rather than a pinned install spec. This is not hidden, but it is less precise than referencing the skill's own installed path.
SKILL_DIR="$(dirname "$(find ~/.workbuddy/skills -name 'smart_extract.py' 2>/dev/null | head -1)")"
Prefer an explicit path to this skill's bundled script and pinned dependency versions; verify the resolved skill directory before running.
Frames may include faces, subtitles, business material, or other sensitive content from the video.
The workflow reads extracted image frames into a multimodal model for analysis. This is expected for video-frame analysis, but it means frame contents may be exposed to the selected model provider or model session.
必须先切换到多模态模型（支持图片输入）... GLM-5v-Turbo 或 GPT-4o ... 用 Read 工具读取 3-4 张帧图片
Use the skill only with videos you are comfortable sharing with the selected multimodal model provider, and redact sensitive content when needed.