Screen Recording for Claude Code

Screen recording for Claude Code, in a form it can read. Record the bug, circle it and say what is wrong. Claude Code gets your words, marks and frames, and fixes it with your code open.

  1. 1Connect Claude Code (once per machine)
    claude mcp add --transport http --scope user vidmatic https://api.vidmatic.ai/mcp

    No key in it: run /mcp, choose vidmatic, then Authenticate, and sign in from your browser.

  2. 2Then prove it works on your app
    /mcp__vidmatic__record --init

    Records a few seconds of your app with one spoken line, puts it in your Inbox and prints a public link. About three minutes on a warm machine.

Loading demo…
Connect Claude Code once, circle a wrong price on screen, and watch Claude Code read the recording, fix it and send back a narrated demo.

Watch on its own page

Claude Code reads text and images, so an mp4 on its own gives it nothing to work with. Vidmatic turns your recording into a timed transcript, every mark you drew with the second you drew it, and an image of the screen at each mark, served by a hosted MCP server.

How it works

1

Connect once

One claude mcp add line and a browser sign-in. No restart.

2

Record and circle the bug

Talk while you record, and draw on whatever is wrong.

3

Claude Code reads it

Your transcript, marks and frames arrive over MCP.

4

The fix comes back with proof

One GitHub issue, the fix, and a narrated demo in your Inbox.

  • Every workflow is a slash command, with nothing copied into your repo
  • No ffmpeg and no local model
  • Your code stays on your machine, with your agent

How screen recording for Claude Code works

Vidmatic holds the video and Claude Code holds the code. The recording reaches Claude Code as text and images.

Record the bug and circle it

Record with the Vidmatic Chrome extension and talk the way you would to a teammate. At the problem, pick up the pen and circle the wrong button or write the right value by hand. Each mark is saved with the second you drew it. For a desktop app, press New Recording in the Vidmatic app and pick any window; there is no pen there, so say which element you mean.

Hand it to Claude Code

Every recording has a Give this to your agent panel. Copy its line into Claude Code and it runs /mcp__vidmatic__workstream_from_video on that recording. It calls get_recording_intel, opens get_annotation_image for each mark, searches your repository and files one GitHub issue with a row per ask, each with its moment, its mark and file and line anchors.

Already have an issue that started from a recording? /mcp__vidmatic__issue_intel with the issue writes the requirements list back to it.

Watch it work, then get proof

As Claude Code plans, builds, reviews and ships, it reports each stage to Agent Workstreams. Driving a session yourself? /mcp__vidmatic__workstream mirrors it there too.

When the change is in, /mcp__vidmatic__record drives your running app in a headless browser, narrates what changed and puts the video in your Inbox. For a fix, /mcp__vidmatic__before_after records the bug today and the fix once it lands, on one timeline.

Why it fits Claude Code

Vidmatic is a hosted MCP server at https://api.vidmatic.ai/mcp, built first for Claude Code. Claude Code and Cursor are the clients we verify end to end; the page on screen recording for AI coding agents, linked below, covers the others.

How it compares, as of October 2026

Clipy (read October 11, 2026) turns each recording into a brief with extracted requests, key moments and a draft plan, and captures console errors. Pro is $9 a month after a 7-day trial.

ThinkRun (read October 18, 2026) is a Chrome extension that records clicks, console, network and narration for a local MCP server. Pro is $19 a month.

Jam (read October 18, 2026) serves console logs, network requests, a transcript and frames from a hosted MCP server. Loom (same date) reaches agents through the Atlassian MCP server, which reads transcripts and metadata.

Vidmatic does not capture console or network data. It keeps what you pointed at, as timed marks with images, and closes the loop with an issue, live progress and before and after proof. The comparison posts linked below have the full tables and sources.

Questions

How much does Vidmatic cost with Claude Code?
The Free plan is free forever and holds 25 videos of up to three hours each, agent hand-off included. Pro is $25 a seat a month for unlimited recordings. Your Claude Code plan is separate.
Are my recordings private when Claude Code reads them?
Yes. Recordings stay private to your account until you share them, and Claude Code signs in as you, so it reads only what you can open. Demo videos it records follow your account's default in Settings, which is public unless you change it.
Does Vidmatic read my terminal or my source code?
No. Claude Code reads your repository on your machine. Vidmatic receives only what the agent sends on purpose: issue text, stage summaries and the demo videos it records of your running app.
Does it work offline?
No. Vidmatic is a hosted MCP server, so Claude Code needs a connection to read a recording or upload a demo.
Can my team use it with Claude Code?
Yes. Share a recording and the Give this to your agent panel on it copies a line with --token, so a teammate's Claude Code reads that one recording. Agent Workstreams shows your own account's work today. Pro is billed per seat.

Related use cases

  • Screen Recording for AI Coding Agents

    Screen recording for AI coding agents: record your screen, circle it and say it. Claude Code or Cursor reads the transcript, marks and frames over MCP.

  • Screen Recording to GitHub Issue

    Turn a screen recording into a GitHub issue your agent can build: numbered requirements tied to the moment you said them, grounded in your code.

  • AI Agent Proof-of-Work Demos

    Proof of work for AI coding agents: when your agent says done, Claude Code records your running app and narrates the change, so you can see it before you merge.

Give Claude Code Your Screen

Connect once, record the bug, and hand Claude Code the recording.

Explore more Vidmatic use cases

Docs: Quickstart