Let your assistant do the boring half of video.

What MCP actually is

MCP stands for Model Context Protocol. Strip the name away and it is simple: a standard way for an AI assistant to use a product the way a person would, after you have signed in and approved what it may touch. Your assistant is not guessing at a screenshot of Rekkko. It is asking Rekkko directly, as you.

So instead of opening the editor, finding the recording, scrubbing for the pauses and cutting them, you type one sentence into the assistant you already have open. The work happens in your Rekkko account, and you see the result next time you open your library.

It is free, on every plan, including the free one. There is nothing to install and no key to paste. You sign in with the account you already use and approve an access screen that spells out exactly what is being granted.

Things people actually ask for

“Clean up the ums in my latest recording.” “Cut every silence longer than two seconds.” “Trim the first eight seconds off, I was still setting up.” “Remove the bit where I said the wrong price.”

“Which recording got the most views this week?” “How did my recordings do this month?” “Find the one where I talked about pricing.” “Summarise the comments on my last three videos.”

“Put a password on that link.” “Make it expire on Friday.” “Switch that link off.” “Move everything with test in the title into a folder called Scratch.” “Undo everything I changed on that recording.”

The assistant shows you what it is about to do before it changes anything, and for anything destructive or anything that changes who can watch, it has to ask you first. That is built into the tools, not left to the assistant's manners.

What it cannot do, honestly

It cannot hand you the finished edited video file. Rekkko builds edited video inside your own browser, using your machine rather than a render farm, which is why editing is free and fast. An assistant can set up every cut, but producing the new file is still one click in the editor. I would rather say that plainly than pretend otherwise.

It cannot record for you either. Recording needs a real browser, a real screen and your permission, so pressing record stays a human job.

It cannot reach anyone else's recordings, not through a share link and not by asking nicely. It cannot change your plan, your billing or your password, and it cannot read a link password you have set, only replace or remove one.

Why I built it this way

The tedious part of video is not recording. It is the ten minutes afterwards, trimming the dead air, cutting the ums, finding the recording from three weeks ago where you explained the thing again. That is exactly the kind of work an assistant is good at, and none of it needs a person watching a progress bar.

The part I was careful about is trust. Reading is cheap and reversible, so reading tools are unrestricted. Changing a recording, changing who can watch it, or deleting anything requires you to say so first, and every edit except deletion can be undone, because the original recording is always kept.