# Check Video > Extract frames from a screencast video in ~/ビデオ/スクリーンキャスト/, analyze the content, and execute instructions. Use when the user wants to verify UI changes, debug visual issues, or document features from recorded screencasts. Use when this capability is needed. - Skill: `tomevault-io/check-video` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/check-video` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/check-video/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/check-video --- # Check Video Screencast Extract frames from screencast videos, analyze visual content, and execute instructions based on the video. ## When This Skill is Invoked Claude uses this skill when: - User runs `/check-video [N] ` - User wants to analyze a recorded screencast - User needs to verify UI changes shown in a video - User wants to document features from a screen recording ## Usage ``` /check-video [N] ``` - `N` (optional): Which video to use (1 = latest, 2 = second latest, etc.). Default: 1 - `instruction`: What to do after analyzing the video content ## Process ### 1. Locate the Video File Find the latest (or N-th latest) video file in `~/ビデオ/スクリーンキャスト/`: ```bash # Get the N-th latest video file (default N=1) VIDEO_FILE=$(ls -t ~/ビデオ/スクリーンキャスト/*.{mp4,webm,mkv,avi,mov} 2>/dev/null | sed -n '${N}p') # Example for latest (N=1): VIDEO_FILE=$(ls -t ~/ビデオ/スクリーンキャスト/*.{mp4,webm,mkv,avi,mov} 2>/dev/null | head -1) ``` ### 2. Extract Frames with ffmpeg Extract key frames at regular intervals (e.g., every 3 seconds): ```bash # Create temporary directory for frames FRAME_DIR=$(mktemp -d) # Extract frames every 3 seconds ffmpeg -i "$VIDEO_FILE" -vf "fps=1/3" "$FRAME_DIR/frame_%04d.png" # Or extract specific number of frames evenly distributed DURATION=$(ffmpeg -i "$VIDEO_FILE" 2>&1 | grep "Duration" | awk '{print $2}' | tr -d ,) # Calculate interval for N frames (e.g., 10 frames) ffmpeg -i "$VIDEO_FILE" -vf "select='not(mod(n\,30))'" -vsync vfr "$FRAME_DIR/frame_%04d.png" ``` ### 3. Read and Analyze Frames Use the Read tool to view each extracted frame: ```bash # List all extracted frames ls "$FRAME_DIR"/*.png ``` Then read each frame with the Read tool to analyze the visual content: - UI elements and their states - Text content visible in the UI - Error messages or dialogs - Visual bugs or layout issues - Feature demonstrations ### 4. Execute the Instruction Based on the analyzed video content, execute the user's instruction: - Fix UI bugs identified in the video - Update documentation with screenshots from the video - Verify that features work as shown - Implement missing features observed in the video - Compare actual behavior with expected behavior ### 5. Cleanup ```bash # Remove temporary frames rm -rf "$FRAME_DIR" ``` ## Frame Extraction Strategies ### Strategy 1: Time-based (Every N seconds) ```bash # Every 3 seconds ffmpeg -i "$VIDEO_FILE" -vf "fps=1/3" "$FRAME_DIR/frame_%04d.png" ``` ### Strategy 2: Fixed Number of Frames ```bash # Extract exactly 10 evenly-distributed frames ffmpeg -i "$VIDEO_FILE" -vf "select='not(mod(n\,$(ffprobe -v error -count_frames -select_streams v:0 -show_entries stream=nb_read_frames -of default=nokey=1:noprint_wrappers=1 "$VIDEO_FILE" | awk '{print int($1/10)}'))'" -vsync vfr "$FRAME_DIR/frame_%04d.png" ``` ### Strategy 3: Scene Changes (Key Frames Only) ```bash # Extract frames only when scene changes significantly ffmpeg -i "$VIDEO_FILE" -vf "select='gt(scene,0.3)'" -vsync vfr "$FRAME_DIR/frame_%04d.png" ``` ## Common Use Cases ### Use Case 1: UI Bug Verification ``` /check-video 1 Verify if the alignment issue in PhotoTags tab is fixed ``` ### Use Case 2: Documentation ``` /check-video 1 Create screenshots for the S3 backup feature documentation ``` ### Use Case 3: Feature Implementation ``` /check-video 1 Implement the emoji icons shown in the maintenance tab ``` ### Use Case 4: Regression Testing ``` /check-video 2 Compare this video with the previous one to check for regressions ``` ## Tips 1. **Frame Rate**: Adjust fps based on video length - Short videos (< 1 min): Extract every 2 seconds (`fps=1/2`) - Medium videos (1-5 min): Extract every 3-5 seconds (`fps=1/3` or `fps=1/5`) - Long videos (> 5 min): Extract every 10 seconds (`fps=1/10`) 2. **Video Format**: ffmpeg supports most formats (mp4, webm, mkv, avi, mov) 3. **Frame Quality**: Use PNG for lossless quality when analyzing UI details 4. **Focus on Key Moments**: Use scene detection to capture only important changes 5. **Temporary Storage**: Always use `mktemp -d` for frame storage and clean up afterwards ## Error Handling - **Video not found**: Verify the path `~/ビデオ/スクリーンキャスト/` exists and contains video files - **ffmpeg not installed**: Install with `sudo apt install ffmpeg` (Linux) or `brew install ffmpeg` (Mac) - **Permission errors**: Check file permissions on the video directory - **Out of disk space**: Clean up old frames from previous runs ## Integration with PhotoClove Development This skill is particularly useful for: - Verifying UI alignment and styling issues - Documenting new features with visual examples - Reproducing bugs shown in screen recordings - Comparing before/after states during refactoring - Creating visual test cases for regression testing --- > Converted and distributed by [TomeVault](https://tomevault.io/claim/ktat) — claim your Tome and manage your conversions.