How to Edit YouTube Videos with GPT6 Astra: A Complete Workflow

Watching a YouTube video is fun. Editing one can mean hours of sorting recordings, cutting repeated takes, finding supporting shots, and adjusting text and audio before it’s ready to upload.

Connected to a video editor through MCP ChatGPT can help you take a lot of grunt out of this process by editing videos for you. With GPT6 Astra selected in ChatGPT, you can describe the edit you want, and it uses the editor’s tools to assemble footage, revise cuts, add supporting visuals, and finish the timeline.

If you’re looking for a free option, invideo Editor can be a great choice. Its MCP connection gives your assistant access to the editing controls you would otherwise operate manually. The work happens on a professional-grade timeline that you can also adjust yourself.

Using invideo Editor’s editing tools through MCP is free and does not consume invideo credits. Your requests use your ChatGPT allowance under your existing plan. If you ask it to generate new images, video, or audio, that generation uses invideo credits.

We’ll use a cooking tutorial on making restaurant-style mushroom risotto as our example. With presenter recordings, cooking close-ups, and footage of the finished dish, we’ll build a video that makes the technique easy to follow without making viewers sit through every minute of stirring.

1. Connect GPT6 Astra to your video editor

Start a ChatGPT conversation with GPT6 Astra selected. Open invideo’s connection page, choose ChatGPT, and follow the setup instructions for your account. Sign in to invideo when prompted and make sure the connection is available in the conversation you’ll use for editing.

You can also connect ChatGPT to another video editor that offers a compatible MCP server. Follow that editor’s connection instructions, as the setup process and available editing capabilities may differ.

If you already have a project, share its link or identify it by name. Otherwise, ask the assistant to create one and upload your recordings to it.

Keep the project open in invideo Editor so you can watch the timeline take shape and review each stage.

2. Explain what viewers should get from the video

Before asking for cuts, explain the video’s purpose. Include the intended audience, the main takeaway, an approximate duration, and any material that must stay.

For our example:

I’m making a six-to-eight-minute YouTube tutorial on cooking restaurant-style mushroom risotto at home. It’s for home cooks who want an impressive dinner but find risotto intimidating. The main takeaway is how to get the texture right and recognise when it’s ready. I’ve uploaded presenter recordings, cooking close-ups, and shots of the finished dish. We’ll shape the edit so the technique is easy to follow.

This gives the assistant context for the decisions that follow, before you ask it to assemble anything. A duration target helps guide the edit, but it should not force the removal of a necessary instruction.

If you have a script or outline, include it as a reference. Let the assistant know whether it should follow that order closely or reorganise the recorded explanation.

3. Build the rough cut

Ask GPT6 Astra to review the recordings and assemble the main sequence. You don’t need to watch everything first and mark each false start yourself.

Build a rough cut from the presenter recordings and cooking demonstrations. Arrange it around preparing the mushrooms, toasting the rice, adding stock, and finishing the risotto. Choose complete takes and remove false starts and repeated explanations. Shorten repetitive stirring, but keep the demonstrations that show how the texture changes. Leave music and graphics for the next pass.

Open the assembled timeline and watch it for structure. Does each section deliver useful information? Are instructions complete? Does the order make sense to someone following along?

For this tutorial, the viewer needs to understand what happens between adding stock and finishing the dish. You can shorten the wait while keeping enough footage to show the change in consistency.

If there are minor changes needed, you can also make adjustments manually on the invideo Editor timeline instead of writing a long prompt. If you want a demonstration to stay on screen longer, adjust the clip’s trim; if two explanations would work better in a different order, move them yourself. The assistant rereads the timeline before subsequent edits, so you can continue giving instructions from the version you’ve just adjusted.

For more detail on this stage, see our guide to building a rough cut with AI.

4. Refine the edit with feedback

Once you’ve watched the rough cut, tell GPT6 Astra what you want changed. You can ask it to strengthen the opening, shorten a repetitive section, or give an important demonstration more time.

Identify the relevant moment using a section name or shot description. For precise edits, include timestamps from the current timeline so the assistant knows exactly where to work. Explain both the change and anything that should stay untouched.

For our tutorial:

Start with the finished plating shot and the line explaining that we’re making restaurant-style mushroom risotto at home. Move into the preparation without repeating the introduction. In the stock section, shorten the stirring footage between explanations, but keep the full demonstration of how to recognise when the rice needs more liquid. Leave the mushroom-browning section unchanged.

For a smaller adjustment, you could give a timestamped instruction such as:

Between 03:20 and 03:35, keep the close-up of the rice on screen instead of cutting back to the presenter. Leave the dialogue unchanged.

Use the timings from your own edit. This gives the assistant a clear location and scope for the change. You can group related changes in one request, then watch the revised sequence before moving on.

See Luuk Alleman request a different opening after making manual adjustments to his edit.

Review the transitions between sections too. Removing every pause can make an instructional video harder to follow, particularly when the viewer needs time to see a texture change or understand a movement. If a section feels rushed, tell the assistant which action or explanation needs more room.

5. Add B-roll footage to make the explanation clear

With the main sequence in place, add supporting shots that show what the presenter is explaining. B-roll lets viewers see the details while the dialogue continues underneath.

For this tutorial, a close-up can show how the mushrooms should look after browning. An overhead shot can make the stock additions easier to follow. A shot of the risotto moving in the pan can illustrate the finished consistency.

Ask for the specific shot and explain where it belongs:

Find the close-up where I drag the spoon through the rice just before adding more stock. Use it as B-roll over the explanation of when the next ladle should go in, keeping the presenter’s dialogue underneath. Hold the shot long enough to see the texture, then return to the presenter.

The assistant can search your footage for matching moments and place the relevant selection on the timeline. You describe what the edit needs instead of searching through every recording yourself.

For a visual example, see Chris Ajtony’s YouTube edit with screen-recording B-roll layered over his talking-head footage.

If a section needs material you haven’t filmed, you can ask it to source stock footage or generate supporting B-roll. For example, a brief shot of mushrooms growing could support an introduction to the ingredient. Use your actual cooking footage to demonstrate the technique and finished result.

6. Add titles, captions, and useful graphics

Use text and graphics to make the recipe easier to follow: ingredient labels can show quantities as they’re mentioned, section titles can mark each stage, and visual cues can draw attention to the texture viewers should look for.

Tell GPT6 Astra what to show, when it should appear, and where it should sit:

Using my recipe notes, show each ingredient’s name and quantity when it is first introduced. Add short title cards for “Preparation,” “Getting the Texture Right,” and “Plating & Serving” at the start of those sections. When the presenter explains the cue for adding more stock, add a brief label beside the close-up to reinforce that instruction. Keep labels clear of the pan and the presenter’s hands.

See WealthWise use on-screen emphasis text to reinforce figures mentioned in the dialogue.

You can also ask for subtitles and specify their styling and placement:

Add readable subtitles throughout. Keep them away from the lower part of the frame when it contains the pan or ingredient quantities. Check ingredient names against the spellings in my recipe notes.

If you want an animated title, describe the movement or share a reference. The assistant can build it with editable keyframe controls inside invideo Editor, so you can refine the motion and timing yourself or request further changes.

7. Build the audio around the video

Tell GPT6 Astra how music, natural sound, and sound effects should work together. For this tutorial, background music can support the cooking sequence, the recorded sizzling can bring the close-ups to life, and subtle sound effects can mark the title cards between sections.

Find a warm, understated background track and add it to the timeline. Bring it forward during the opening plating shot, then lower it when the explanation begins. Preserve the sound of the mushrooms sizzling and the stock going into the pan. Fade the music out at the end.

The assistant can find and add background music within the editor, keeping music selection and placement in the same workflow.

You can then ask it to add sound effects timed to the edit:

Add a soft, short whoosh as each section title card appears. Use the same treatment for “Preparation,” “Getting the Texture Right,” and “Plating & Serving,” and let it finish before the presenter starts speaking. Keep it subtle so the transition feels distinct without interrupting the tutorial.

You can see generated sound effects arranged on separate audio tracks in LunaAi’s demonstration.

Follow with targeted cleanup where needed:

Reduce the background hum in the presenter recordings and smooth the dialogue levels between takes. Keep the cooking sounds in the demonstration footage.

Listen through the joins between recordings and the transitions between sections. Check that the music and effects support those moments without covering dialogue or making the change in sound feel abrupt.

8. Review the finished edit and create chapters

Watch the full video with the sound, captions, and graphics in place. Check that the opening matches what the video delivers, demonstrations remain understandable, and the text is accurate.

Once you are happy with the edit, ask the assistant to review the sections and create chapter markers:

Review the completed timeline and add chapter markers based on the content of each section. Give me matching timestamped chapter titles for the YouTube description.

If you change the edit after this, ask it to update the chapter timestamps and markers to match.

9. Export the finished video

Identify the approved timeline and ask for a rendered video:

Export the approved “Mushroom Risotto Tutorial” timeline as a finished video for YouTube.

invideo Editor supports a finished H.264 MP4 export.

You can also export an editable project for Premiere Pro, Final Cut Pro, or DaVinci Resolve if you want to continue working in another application.

Review the exported file before uploading it. Check the beginning and ending, listen to the audio, and review any captions or graphics changed during the final pass. Once everything looks and sounds right, upload the finished video to YouTube and add the timestamped chapter list from the final edit to its description.

Frequently asked questions

Can GPT6 Astra edit a YouTube video I’ve already started?

Yes. Through a compatible MCP connection, it can work on an existing editor project. In invideo Editor, share the project and identify the timeline, then explain what you want changed and which parts should remain untouched.

Is editing YouTube videos with GPT6 Astra free?

Using invideo Editor’s editing tools through MCP is free and does not consume invideo credits. Your requests count towards your ChatGPT usage allowance, and any ChatGPT subscription costs remain separate. Generating new images, video, or audio uses invideo credits.

Do I need to write one prompt for the entire video?

No. Explain the overall video first, then direct the work in stages. Reviewing the rough cut before adding supporting shots, graphics, and sound gives you room to make decisions as the edit develops.

Can I keep editing manually?

Yes. The assistant works on the same invideo Editor timeline you can adjust yourself. You can change cuts, move clips, refine titles, or adjust audio, then continue directing the assistant from the updated edit.

Can I try different versions of my YouTube edit without losing the original?

Yes. Ask GPT6 Astra to duplicate your edit onto another timeline in invideo Editor before making changes. You can try a different opening, pacing, or shorter cut, compare the versions, and keep the original intact.