Getting Started with Tella MCP 🚀

In this video, I'm gonna show you how you can set up and start using the Tella MCP to edit videos.

first thing you need to do is go to our docs.

Go to docs.tella.com, then click on Developers, then click on Model Context Protocol, and then you'll get some setup and authentication instructions for whichever AI tool you're using.

I'm using Claude Code, and so that's what I'm gonna be demoing today.

To start, just copy this command here, paste it into your terminal before you, even start running Claude Code, and that will add the config, to your, to your config file locally.

and then you'll need to authenticate.

So, if I jump into the terminal, I've already got the, the, the local config.

But then if I just run /MCP, you can see that I've got user MCP is Tella, and it says "Connected." If this wasn't connected, you would see that it says that it needs to be authenticated.

So then just hit Tella, and then that will prompt you to sign in to your Tella account on the web.

So, let's just open up the, the MCP server.

We can explore it like this.

We can view the tools, we can authenticate, clear authentication, reconnect, disable.

let's just view the tools, and then this is gonna let us navigate through all the different options we have.

if you go back into the docs, then you can sort of cross-check all of that stuff in here.

So you can explore what the MCP can do in the docs or directly from your command line.

The other thing you can do is you can just say, "Hey, tell me what edits I can make to videos using the..." If you're not super technical or super handy with the command line, which is the case for me, then I just like to ask, Claude to go and do these things for me, which it usually can.

so here's a good summary of the types of edits that we can do with the MCP.

Cutting clips, removing filler words.

We've got a bunch of visual edits like zooms and layouts.

you can add B-roll and things like that.

You can manage your videos.

The update that we're announcing today is around all of these editing tools, and so that's what I'm gonna show you, right now.

So let's go and grab a video and start making some edits.

this is a video that we're gonna edit, thirteen minutes long, screen and camera recording, one clip.

and so we can just copy the URL to, to our clipboard, and then we can jump back into the terminal and say, "Let's edit this.

Start by removing all the silences." and, that is then going to go through the MCP, figure out what tools it needs to use to make the changes we've asked for.

and in this case, because we're using the silence removal tool, it's probably gonna come back to us, with a, a request of how we wanna configure the silence removal.

Do we want it to remove, just a little bit of silence, or do we want it to remove a lot of silence and be a very aggressive, edit?

So I'll see you back here in a moment when it's finished thinking and, we can, wrap up this edit.

it's done.

in this case, it didn't actually ask us for any follow-up, information.

It just went and removed the silences for us, which is fine.

it cut three minutes of, dead air, which is great.

That really cuts down our, our video and just shows how much, unneeded space we have in there.

and so now what we're gonna do is we're gonna jump into the editor to sort of verify what it's done and make sure all of that looks good.

So here in the cut tool, I've opened that up, and you can see on the timeline that it's removed all of the silences.

You can just see the areas with the waveform remaining.

And, if we want to, we can manually adjust this to clean up anything that maybe it did that we're not so happy with or we want to, to tweak.

But other than that, we're good to go.

That is our first edit done via the MCP.

Next thing we're gonna do is add some zooms to the MCP back in the terminal, you can actually see that it has prompted us to, decide if we want to add some, some zooms.

so we can just say, "Yeah, add some zooms to the video," and it's gonna go ahead and get to work on this change.

there are two different types of zooms.

You can just do the standard zoom and pick a focus point and a zoom level, or there are tracked zooms where it follows the movement of your cursor, and, and has a zoom level as well.

So, let's see what it comes back with, and which zooms it wants to add.

All right, it's done.

Added 11 zooms.

it went through the transcript to try and figure out, where some good places to add the zooms are.

So you can see it wanted to check if there were existing zooms and also check the transcript to, to figure out where to place them.

and, each of them is about six to nine seconds over substantial spoken segments.

Cool.

Okie doke.

So, now what we're gonna do is we're gonna jump into the editor, take another look, see what it's done and, and make sure that looks all good.

So let's switch to the zoom tool.

And here you can see we've got the first zoom added.

It's a auto zoom, so it tracks the movement of the cursor.

And if we zoom out now, you can see that it's added, some zooms throughout the full duration of the video.

And, as a quick eyeball, that looks okay.

Just like the trims and any other edit you make by the MCP, you can manually adjust it or tweak it here in the UI.

So you can use the agent as a starting point, and then you get to be kind of like a final, a final tweaker of, of the changes that it makes.

Now, editing videos via the MCP like this is fun, but it's actually more efficient to work in the UI, in the actual Tella editor when you want to make a change and review it visually and then iterate from there.

The power of the MCP is when you want to automate editing, so you don't have to do anything.

It can run in the background or when you want to do a lot of things at once.

And the best demonstration of this is with a skill.

Now, a skill is something that you can use in, Claude Code or Codex, which is basically a set of instructions that's saved in the file that you can then reference, and your agent will be able to follow all the instructions in the skill without you having to write them out every single time.

And the nice thing about them is that you can write really, really detailed instructions with lots of, logic around how the model, how the agent should, in this case, edit a video.

So I've made a very simple skill already, which, combines some common edits that I like to make to my videos, that builds off the couple of edits that we have already made.

So I am gonna say, "Use the clean edit skill," that's what the skill is called, "to edit this video." And I'll just paste in the link there.

This is the same video that we've been working on already, but I've reset all the edits we made before, and then the clean edit skill is going to, edit this again from scratch.

So when we type this in, it's gonna go and use that, skill and, and then get to work on that video.

So, I'll see you back here once it has finished its work, and we'll, review what it's done.

it's completed the edit.

it just did it in one go.

We didn't need to add, any extra info, which is what we want.

We wanna be able to, like, make a skill or an automation like this that can just run in the background, and we can come and check it when it's done or we're ready.

So you can see-- let's go and see what, what it's done.

loaded the video, inspected it, made sure it's got, all of its core info.

then it started working on, the cuts, which were to remove all the silences.

then if we jump down here, this is where it started adding the zooms.

so it has gone and looked for places to add zooms based on the transcript, and then decided to add thirteen, making sure that nothing overlapped.

and then it wrapped up the edit with a intro, of a camera-only shot for the fif- first five seconds and then-- and a, a same, layout of just the camera for the, the final five seconds.

and then kind of wraps it up here with a summary.

So let's jump back into the editor and then recap, all these changes visually.

So let's start by checking out the cut timeline, and here you can see it's removed all the silences.

Similar looking result to when we did this, edit individually.

Then if we jump to the zoom timeline, you can see it has added some zooms for us here as well, which is looking good.

And then the, thing you'll-- you might already notice is that instead of showing our screen in the first frame, it shows our camera.

and that's because it has added a camera-only layout for, for the intro, which then transitions into the screen plus camera one, and then we can just jump to the end of the, video to verify that we have the same thing for the outro, which is looking good there.

so we could make some adjustments or tweaks here, but, overall, that's looking really good.

That's a one-shot edit of a whole video using the Tella MCP.

From here, I'd recommend going deeper into the docs and getting a good understanding of all the different tools, that you can use.

I really just scratched the, the surface.

You can also just chat with your agent and figure out, what ki-- type of edits and, and things you can do like that, and then just experiment.

Try some stuff, have longer conversations with your agent around, the different types of, of edits you can make, and, and iterate on those edits.

And then when you find stuff that works and you get really nice results, save that as a skill you can then reuse, again later.

Thanks for tuning in.

As always, let us know what you think and, we'll talk again soon.

Bye for now.