agent-harness-design
Designing agent harnesses and tool systems — risk taxonomy for tools, permission decisions, draft/commit pattern, structured tool…
Video post-production rules: audio mastering, color, captions, platform export. Use when: 'add music', 'add voiceover', 'export for tiktok', 'add captions', 'color grade', 'audio levels', 'master audio', 'export settings', 'platform requirements'. Covers FFmpeg patterns, audio
$ npx -y skills add AnastasiyaW/codex-claude-code-config --skill video-post-production --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/video-post-productionContext preview
The summary Claude sees to decide when to auto-load this skill.
Video post-production rules: audio mastering, color, captions, platform export. Use when: 'add music', 'add voiceover', 'export for tiktok', 'add captions', 'color grade', 'audio levels', 'master audio', 'export settings', 'platform requirements'. Covers FFmpeg patterns, audio
name: video-post-production description: "Video post-production rules: audio mastering, color, captions, platform export. Use when: 'add music', 'add voiceover', 'export for tiktok', 'add captions', 'color grade', 'audio levels', 'master audio', 'export settings', 'platform requirements'. Covers FFmpeg patterns, audio chain, subtitle standards, and platform-specific export configs. Do NOT use to generate or animate the video frames/scenes in code (use remotion-production-guide), to write the script (use video-narrative-arc), or to evaluate script quality (use script-evaluator); this is the finishing stage applied to an already-rendered video."
Rules and FFmpeg patterns for finishing videos: audio mastering, captions, color, and platform-specific export.
| Track | Level | Notes | |-------|-------|-------| | Voiceover | -12 to -10 dB | Primary audio, clear and upfront | | Music (under VO) | -32 to -24 dB | Background, should not compete | | Music (solo, no VO) | -16 to -12 dB | Can be louder when alone | | Sound effects | -18 to -14 dB | Accent, not distraction | | Silence floor | -60 dB | True silence feels unnatural, keep room tone |
| Mood | BPM Range | Use Case | |------|-----------|----------| | Calm, reflective | 60-80 | Luxury brands, testimonials | | Confident, steady | 80-100 | Corporate, explainers | | Moderate energy | 100-120 | Product demos, features | | Energetic | 120-140 | Launch videos, social ads | | High energy | 140+ | Montage, fast-cut sequences |
**Rule:** Match music tempo to edit pace. Cut on beats for professional feel.
# Full mastering chain: highpass + EQ + compression + loudness ffmpeg -i input.mp4 -af " highpass=f=80, equalizer=f=3000:t=q:w=1.5:g=3, acompressor=threshold=-20dB:ratio=4:attack=5:release=50, loudnorm=I=-16:TP=-1.5:LRA=11 " -c:v copy output.mp4
| Parameter | Value | |-----------|-------| | Max characters per line | 32-42 | | Max 2 lines simultaneously | Always | | Display duration | 1.5-7 seconds per caption | | Min display | 1.5 seconds (even for short text) | | Font size | 48-72px (for 1080p) | | Font | Sans-serif, bold weight | | Position | Bottom center, 10% from bottom edge | | Background | Semi-transparent black (#000000AA) |
# Basic caption burn-in ffmpeg -i input.mp4 -vf "subtitles=subs.srt:force_style='FontSize=24,FontName=Arial,Bold=1,PrimaryColour=&HFFFFFF,OutlineColour=&H000000,Outline=2,MarginV=40'" output.mp4 # Word-by-word highlight (requires ASS format) ffmpeg -i input.mp4 -vf "ass=styled_subs.ass" output.mp4
# Warm punch (indoor/talking head) ffmpeg -i input.mp4 -vf "eq=contrast=1.1:brightness=0.02:saturation=1.15,colorbalance=rs=0.05:gs=-0.02:bs=-0.05" output.mp4 # Cool tech (product/SaaS) ffmpeg -i input.mp4 -vf "eq=contrast=1.05:saturation=0.95,colorbalance=rs=-0.03:gs=0.01:bs=0.05" output.mp4 # High contrast B&W ffmpeg -i input.mp4 -vf "hue=s=0,eq=contrast=1.3:brightness=0.03" output.mp4
ffmpeg -i input.mp4 \ -c:v libx264 -preset slow -crf 18 -pix_fmt yuv420p \ -c:a aac -b:a 192k -ar 48000 \ -movflags +faststart \ youtube_output.mp4
# Resize landscape to vertical with blur background ffmpeg -i input_16x9.mp4 \ -filter_complex "[0:v]scale=1080:1920:force_original_aspect_ratio=decrease,pad=1080:1920:(ow-iw)/2:(oh-ih)/2:black[fg];[0:v]scale=1080:1920,boxblur=20:20[bg];[bg][fg]overlay=(W-w)/2:(H-h)/2" \ -c:v libx264 -preset fast -crf 23 -pix_fmt yuv420p \ -c:a aac -b:a 128k \ tiktok_output.mp4 # Native vertical render from Remotion npx remotion render src/index.ts Comp out/vertical.mp4 --width 1080 --height 1920
# Center crop to square ffmpeg -i input_16x9.mp4 \ -vf "crop=ih:ih:(iw-ih)/2:0,scale=1080:1080" \ -c:v libx264 -crf 23 -pix_fmt yuv420p \ -c:a aac -b:a 128k \ square_output.mp4
# Fast trim (keyframe-based, instant) ffmpeg -ss 00:00:05 -i input.mp4 -t 15 -c copy trimmed.mp4 # Frame-accurate trim (re-encodes) ffmpeg -i input.mp4 -ss 00:00:05 -t 15 -c:v libx264 -crf 18 trimmed.mp4
# Create file list echo "file 'scene1.mp4'" > list.txt echo "file 'scene2.mp4'" >> list.txt echo "file 'scene3.mp4'" >> list.txt # Lossless concat (same codec/resolution) ffmpeg -f concat -safe 0 -i list.txt -c copy output.mp4 # Re-encoded concat (different sources) ffmpeg -f concat -safe 0 -i list.txt -c:v libx264 -crf 18 output.mp4
# Music at -24dB under existing audio ffmpeg -i video.mp4 -i music.mp3 \ -filter_complex "[1:a]volume=-24dB[music];[0:a][music]amix=inputs=2:duration=first" \ -c:v copy output.mp4
# 2x speed (with audio pitch correction) ffmpeg -i input.mp4 -vf "setpts=0.5*PTS" -af "atempo=2.0" fast.mp4 # 0.5x slow motion ffmpeg -i input.mp4 -vf "setpts=2.0*PTS" -af "atempo=0.5" slow.mp4
# High quality GIF with palette ffmpeg -i input.mp4 -vf "fps=15,scale=480:-1:flags=lanczos,split[s0][s1];[s0]palettegen[p];[s1][p]paletteuse" output.gif
Before publishing, verify:
A practical configuration kit for Claude Code, Codex, and other coding agents. It contains architectural principles, enforcement hooks, skills, drop-in rules, starter templates, and dynamic-workflow commands.
Designing agent harnesses and tool systems — risk taxonomy for tools, permission decisions, draft/commit pattern, structured tool…
Use when designing, auditing, refactoring, or explaining an agentic harness for any domain, especially when work must continue…
Практическая инженерия диффузионных моделей: архитектуры, обучение, инференс, оптимизация памяти. Использовать при любых задачах…
Expert prompt engineering for FLUX.2 [klein] image generation and editing model. Use this skill whenever the user wants to create…
Plan or review LoRA and edit-training work specifically for FLUX.2 Klein or Qwen-Image-Edit, including paired datasets,…
Forensic image-to-prompt compiler for image generation models. Use this skill whenever the user wants to: convert/describe an…