What are good prompts for AI video editing?

Good prompts describe the finished result, point to a moment the editor can find, and say what must not change. For example: “Cut pauses longer than half a second, add karaoke captions three words at a time, and reframe to 9:16 for Reels.”

Can I combine several prompts in one message?

Yes. Paste two or three prompts together and the agent orders the steps itself: cuts first, then captions timed to the new cut, then framing and the mix. For long videos, settle the structure first and add style in a second message.

Do these prompts work in other AI video editors?

The plain-English instructions work in any agentic video editor. Preset names such as “karaoke” captions or the “cinematic” grade are Valmera’s, so another tool may call them something else.

Should I use timestamps or describe the moment?

Either works. A spoken phrase is the sturdiest anchor because it still points to the right place after earlier cuts. If you use a timestamp, say whether it comes from the original recording or the edited preview.

How do I fix an edit I don’t like?

Name the problem, where it is, the change you want and what to keep, for example: “The music is too loud from 0:30 to 0:50. Lower it 4 dB and keep everything else.” To step back a whole version, use Previous edit in Studio.

Are these prompts free to use?

Yes. Copy and adapt them freely. Creating a Valmera account and uploading are free; editing requires a subscription.

AI Video Editing Prompts

79 COPY-PASTE PROMPTS · 11 TASKS

Upload your footage to Valmera, paste a prompt, and swap the example details for your own. Every prompt here maps to something Valmera’s editing agent does today, and you can combine several in one message.

Want to write your own? Read how to write AI video editing prompts for the formula and before-and-after rewrites.

How to use these prompts

Upload your video (MP4, MOV, M4V, MKV or WebM, up to 100 GB and 3 hours), plus any logo, music or clips you’ll mention.
Paste a prompt and replace the quoted phrases, timestamps and names with your own.
Watch the preview. The agent renders it, checks it and tells you what it changed.
Revise with a follow-up, then export a full-resolution MP4.

Cleanup 1–7

Cut dead air, filler words and false starts. Cuts land on word boundaries, and anything removed can be restored. Silence removal guide →

Cut every pause longer than half a second, but leave a short breath between sentences.

Remove the ums, uhs and ers. Also cut every “you know.” I say it too much.

I restarted a few sentences. Keep the last complete take of each one and cut the false starts.

Delete everything before I say “let’s get started” and the dead air after my last sentence.

Cut the part where I stop to check my notes, from “hang on” until I say “okay, where was I.”

Remove the tangent that starts at 4:10 and ends when I say “anyway, back to the setup.”

You cut too tightly around 6:20. Restore that section and leave its pauses alone.

Captions and on-screen text 8–15

Word-timed captions burned into the video, styled by preset name (clean, karaoke, podcast, beast, elegant and more), plus titles, callouts and title cards. Captions guide →

Add word-timed captions in the clean preset at the bottom, five words max per caption.

Use karaoke captions with a yellow box on the word being spoken.

Make the captions bigger and all caps in Anton, with a pop animation.

The captions spell my company “Valmira.” Change it to “Valmera” everywhere.

Hide the captions from 1:05 to 1:20 while the full-screen chart is up.

Add a callout that says “Don’t skip this” over the mistake at 1:12.

Cut to a black title card that says “PART 2: THE FIX” for two seconds before the second half.

Put the word “FOCUS” in big letters behind me as I walk through the shot at 0:12.

Text-behind works when a person moves in front of the words within one continuous shot.

Shorts, Reels and TikToks 16–22

Pull one vertical clip out of a longer video, or split a long recording into several shorts. AI clip maker guide →

Find the strongest 45-second moment, cut everything else, and make it 9:16 with captions.

Turn this interview into separate shorts, each one a complete story with a setup and a payoff.

Each short becomes its own project that you edit, review and export.

Start the short on the line “nobody tells you this.” No intro before it.

Keep it under 60 seconds for YouTube Shorts and end right after the punchline.

Show the hook as big text on screen for the first three seconds.

Freeze on my last line for three seconds with “SAVE THIS” over the frame.

Make it TikTok-ready: 9:16, karaoke captions, and audio mastered for phone speakers.

Podcasts and interviews 23–30

The transcript carries speaker labels, so you can refer to the host, the guest or a topic instead of hunting for timestamps. Podcast editing guide →

Clean up this two-person episode: cut long pauses and filler words for both speakers and keep everything else.

Drop the first eight minutes of small talk and open on the first real question.

Find every time the guest talks about pricing and give me the timestamps.

Add lower thirds the first time each person speaks: “Sam Lee · Host” and “Priya Shah · Founder, Northwind.”

Put my uploaded intro music under the first ten seconds, then fade it out under the first question.

I was too close to the mic from 12:00 to 12:45. Lower that range by 3 dB.

Reframe to 9:16 with blurred padding so both of us stay in frame.

Master the whole episode to standard loudness for YouTube.

Reframing and zooms 31–37

Switch to 9:16, 1:1, 4:5 or 16:9 by crop, padding or blurred padding, and add zooms only where they help. Auto reframe guide →

Reframe to 9:16 for TikTok and keep my face in the crop.

Reframe this edit to 1:1 for the Instagram feed.

Make it 4:5 with blurred padding. Don’t crop off the whiteboard.

The vertical crop cuts off my hands when I gesture. Use padding instead of cropping there.

Switch to vertical when I start reading the message, then back to 16:9 when I’m done.

Zoom into the dashboard from 0:24 to 0:41 so the numbers are readable on a phone.

Punch in on the three words I stress the most.

B-roll, logos and inserts 38–44

Stock footage, real photos and your own clips, laid over your voice or spliced into the timeline. B-roll guide →

Add b-roll wherever I describe a place or an object, 3–4 seconds each, with my voice still playing.

When I mention remote work, show a stock clip of someone working on a laptop at home.

When I mention the Eiffel Tower, show a real photo of it for three seconds.

Use the product clips I uploaded as b-roll whenever I talk about the app.

Splice my screen recording in right after I say “here’s how it works.”

Put my logo in the top-right corner at about 8% of the frame width for the whole video.

Show my webcam clip as picture-in-picture in the bottom-left during the demo.

Picture-in-picture clips play silently over your main audio.

Music, sound effects and voiceover 45–52

Bring the music: upload a track, paste a link or name a song. Sound effects are real recordings found on the web. Valmera doesn’t generate music. Music guide →

Put the track I uploaded under the whole video, low and ducked under my voice.

Use the song from this link as the music bed, starting at the chorus: [paste link]

You need the rights to any music you bring.

Start the music when I say “let’s dive in” and fade it out over the last five seconds.

Snap the montage cuts to the beat of the song.

If the song has no steady beat, the agent tells you instead of guessing.

Add a whoosh on each scene change and a camera-shutter sound when the product photo appears.

Remove the background music from my original recording but keep my voice.

Dense mixes can leave traces of the music, so listen to the preview.

Use the audio from the TikTok clip I uploaded as the music.

Lay my voiceover file over the b-roll section from 0:45 and lower everything else under it.

Color, looks and transitions 53–59

Grades apply to the whole video. Effects like grain, vignette or VHS can cover just a time range. Effects docs →

Apply the cinematic grade, then make it slightly warmer.

Use the clean look: simple white captions, no grade, gentle fades.

It’s a bit dark. Brighten it and lift the shadows without washing out the contrast.

Add film grain and a vignette only during the flashback from 2:40 to 3:05.

Use a quick dip-to-black transition at scene changes only, not on every jump cut.

Add a VHS effect over the first five seconds, then go back to normal.

Make it a little sharper. The camera smeared the detail.

Screen recordings and product demos 60–66

Make small interfaces readable: a floating window, a bigger cursor, zooms on clicks and faster loading screens. SaaS demo playbook →

Put the screen recording in a floating window on a dark gradient background.

Make the cursor bigger and steadier. It’s hard to follow.

Follow the cursor with a smooth zoom as it moves from the sidebar to the settings panel.

Speed up the two loading screens to 4x and keep everything else at normal speed.

Zoom in on each click. They happen at 0:08, 0:15 and 0:31.

For your own recordings, give the click times. A click can’t be detected from the picture alone.

Record a demo of https://example.com: click “Pricing,” scroll to the plans, then click “Start free trial.”

Records the public site as a visitor sees it. It won’t type into password or payment fields.

Record https://example.com scrolling from top to bottom and open the video with it.

Privacy and on-screen cleanup 67–72

Blur, pixelate or black out a region, or erase burned-in text and watermarks so the background shows through. Blur guide →

Blur the email address in the top-right corner from 1:02 to 1:19.

Pixelate the customer names in the CRM table for the whole screen recording.

Put a black bar over my phone number at the bottom of the screen.

This clip has old burned-in subtitles. Remove them so I can add new captions.

Erase the app watermark in the bottom-right corner. Remove it, don’t blur it.

Erase the sticky note on the monitor behind me.

Blur and erase regions are fixed rectangles. They don’t follow a moving subject.

Review, revise and export 73–79

Ask the agent to check its own work, explain what changed, and fix what’s off before you export. Export docs →

Render a preview and tell me anything that looks or sounds off.

Check the captions for misspelled names and words that show up late.

Show me the transcript of what’s left after the cuts so I can check nothing repeats.

List every change you made in this version.

The cuts feel rushed. Add back a little breathing room after each sentence and keep everything else.

The music is too loud on phone speakers. Lower it 4 dB and render a new preview.

Export this version as the final MP4 and give me the download link.

For Claude, ChatGPT and other MCP clients. In Studio, press Export.

Paste a Prompt Into the Real Editor

Creating an account and uploading are free. Editing requires a subscription.

Create an account →

See pricing →

Five tips for writing your own

Describe the finished result. “Under 60 seconds for Shorts” works better than “make it shorter.”
Point at something the editor can find. A phrase you said, a timestamp, a speaker, or something visible on screen.
Say what must not change. “Keep the pauses after each instruction” protects the parts you like.
Name the platform. “For Reels” sets the shape, caption size and loudness target in two words.
Fix by describing what you see. “The captions cover the logo at 0:42” beats “try again.”

The prompt-writing guide turns these into a four-part formula, with eight before-and-after brief rewrites.

What no prompt can do in Valmera

These are product limits, not phrasing problems. Here’s what to do instead.

Crossfade between these shots.Use a dip to black or dip to white instead. True crossfades aren’t supported.

Export the captions as an SRT.Captions are burned into the video. There’s no SRT or VTT import or export.

Use my brand font.Pick a bundled font such as Anton, Bebas Neue, Montserrat or Playfair Display. Custom font uploads aren’t supported.

Make the sticker follow my hand.Overlays don’t track motion. Place them where the subject stays put.

Remove the background noise.There’s no audio denoise. Record clean audio, or clean it up before uploading.

Write a soundtrack for this.There’s no AI music generation. Upload a track or paste a link.

Generate a video about remote work.Valmera edits footage you provide. It doesn’t generate video from text.

Post it to my TikTok.Download the MP4 and post it yourself.

Use these prompts in Claude or ChatGPT

Connect Valmera’s MCP server at https://valmera.io/mcp/server and paste the same prompts into Claude, ChatGPT, Cursor or another MCP client. Valmera exposes 143 tools (live server card, October 2026); the assistant calls them itself, renders previews and exports the final MP4. It can also use context already in your chat, like brand notes or last month’s caption settings. Setup: Claude, ChatGPT, tool reference.