Deadspace Remover For TikTok Instagram Reels — Communication skill for Claude Code
Claude Code skill: cut the dead space out of talking-head videos — between sentences only, tight to speech, never clips a word.
How to install Deadspace Remover For TikTok Instagram Reels
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open carrot1445/Deadspace-Remover-for-Tiktok-instagram-reels and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Deadspace Remover For TikTok Instagram Reels does
Claude Code skill: cut the dead space out of talking-head videos — between sentences only, tight to speech, never clips a word. Local ffmpeg + faster-whisper.
Alternatives in Communication
- YouTube Clipper Skill — Download videos, generate semantic chapters, clip segments, translate subtitles to bilingual format, and burn 1.6k ★
- NullMotion — Motion graphics preview and YouTube clip maker: finished ad over HyperFrames drafts, MP4 export, pause cuts, c 582 ★
- Mcptube — Transform YouTube videos into a compounding knowledge base with transcripts, vision analysis, and agentic sear 152 ★
README
tight-cut: a Claude Code skill for cutting dead space
Drop your raw talking-head videos into Claude Code and say "edit these". You get back the same videos with the dead air between sentences removed.
- Tight: each clip starts the instant you start talking and ends the instant you stop, with zero gap between sentences.
- Safe: it never cuts inside a sentence and never clips a word. Soft word endings like the "-ting" in "marketing" are kept, while breaths, mouth clicks and room noise are cut.
- Looks right: it checks the frames so a clip doesn't end on you reaching to stop the recording.
- Local and free: ffmpeg plus faster-whisper. No API keys, and nothing is uploaded.
- Organized: output goes to
finished videos//.
Made for phone selfie videos (TikTok, Reels, Shorts, UGC), where you record several takes back to back and just want the pauses gone.
Install
git clone https://github.com/carrot1445/tight-cut-skill
cp -r tight-cut-skill/tight-cut ~/.claude/skills/ # personal skill, every project
# or: cp -r tight-cut-skill/tight-cut /.claude/skills/
pip install -r ~/.claude/skills/tight-cut/requirements.txt
You also need `ffmpeg` on your PATH (`brew install ffmpeg` / `apt install ffmpeg` / `winget install ffmpeg`). The first run downloads the whisper `small` model (~460 MB, cached after that).
Use
In Claude Code:
edit these: ~/Downloads/clip1.MP4 ~/Downloads/clip2.MP4
or `/tight-cut ~/Downloads/clip1.MP4`. Claude runs the cutter, checks that no words were lost, reads the frame sheet for start/stop-recording reaches, and tells you where the files are.
Or run the script yourself without Claude:
python ~/.claude/skills/tight-cut/scripts/tight_cut.py clip1.MP4 clip2.MP4 --verify
clip1.MP4: 76.06s -> 57.19s, removed 18.87s, 14 clips
-> finished videos/2026-10-14/clip1_tight.mp4
edge sheet -> finished video
Related Skills
Snapcut
An Agent Skill that edits a vertical talking-head take into a finished Reel, TikTok or Short: word-synced capt
Script To Video
Turn a script into a narrated, word-synced explainer video (16:9 YouTube or 9:16 Reels/Shorts). Free local Kok
Sv Auto Learner
SV Auto Learner. Paste a YouTube link in Claude Code or Codex: it reads every word (captions or local Whisper
Sum The Yt
Summarize YouTube videos into Traditional Chinese (zh-TW) — subtitle-first with Whisper fallback, summarized v
Tag Posthog
Tag @posthog in Slack #claude-code-notifier to delegate a task to the PostHog Claude Code agent. Drafts a tigh
Capipaste
Talk to your coding agent about what you see. ⌘⇧S → drag, draw, say what should change, ↩ — the marked-up scre
Related Agents
Editor Videos
Genera videos verticales para TikTok (edits de juegos/pelis y brain rots) con el pipeline local ffmpeg + Polli
Voice Engineer
GAIA voice interaction specialist. Use PROACTIVELY for Whisper ASR, Kokoro TTS, the Talk SDK, speech-to-speech
Dialogue Naturalness Reviewer
Seventh of book-forge's QA reviewers. Asks whether dialogue sounds like people talking rather than like a mode