All recipes
Put zooms and sounds exactly where things happen
A zoom half a second late looks like a mistake. ClipMaster already knows when the camera cuts, who speaks, where the laugh is and where the audio hits. Claude reads that timeline and puts each edit on its second.
The prompt
Paste it into Claude or ChatGPT and change the details to yours.
Look at what happens in clip 2 of my latest video. Zoom in on the laughs, add a flash on the loudest moment and a whoosh at each camera cut. Tell me what you placed, then export it.
What you need
- A ClipMaster account on a paid plan
- The ClipMaster connector added to Claude, ChatGPT or Claude Code
- Tokens for the import and the renders, since every export spends some
Paste this into Claude or ChatGPT as a custom connector
https://api.clipmaster.app/mcpWhat happens
- Claude reads the clip's timeline: camera cuts, shot and face changes, who speaks when, laughter and applause, loud and quiet stretches, beats and the moments the AI scored.
- If the clip's layout hasn't been analysed yet, Claude runs the layout analysis first: it's what finds the camera cuts. Paid plans; it spends no tokens.
- If the clip's audio hasn't been measured yet, it measures it first: that's what finds the loudest moment. Also paid plans; it takes a moment and spends no tokens.
- It places the zooms, the flash and the sound effects on those seconds. Every time in the timeline falls inside the clip as it plays, on the clip's own clock, so nothing needs converting.
- It tells you what went where, then exports the clip and sends you the link.
Tools it uses:
list_projectslist_clipsget_clip_momentsanalyze_clip_layoutmeasure_clip_audioget_clip_analysislist_sound_effectsadd_clip_overlaystrigger_exportget_export_statusTips
- Ask "what happens in clip 2?" on its own first to see the timeline before anything is placed.
- Want the zoom to follow the speaker's face? Claude runs the layout analysis first, which finds the faces.
- Don't like a placement? Ask Claude to move or remove it.