LaunchReel AI / README to video
README to video, with the code still readable
Paste the link to your repository and its README becomes the script: the title and description open the film, the feature list becomes the middle shots, and commands and numbers are set as real text. You approve the brief and the storyboard before anything renders, and the film comes out as a 16:9 MP4 with burned-in captions and an .srt file.
No email and no card to start. You see the brief and the storyboard first; we ask for an email only when you want the preview rendered.
What is free, what costs credits
- Free, no email: the brief and the storyboard, written from what you paste. One storyboard rewrite is free too.
- Free once per verified email: a watermarked 720p preview of the film (and a few per network per day).
- Credits from one-time packs, from $9: 1 to change a shot, 10 for a clean 1080p export, 20 for 4K, 5 to start another product. See pricing.
What the film is today
- 16:9 only, 1920×1080 or 4K. A 9:16 vertical cut is not available yet.
- Silent: captions are burned in and you also get an .srt file. No voiceover or music yet.
- MP4 files. No GIF export.
Why the text stays legible
Video generators that paint every frame tend to garble small text: a command loses a character, a number changes between frames. LaunchReel AI does not paint frames. Each shot is a motion-design scene that places your words as real type, so the command in the film is the command in your README, character for character, and you proof it on the storyboard before it renders. The Claude page shows what one of those scene files looks like.
What is read, and in which language the film is written
We read the repository page as the code host renders it, so the README's headings, lists, code blocks, tables and images are all available to the brief, plus up to three pages linked from it (a guide, the docs). Files that the page does not show or link are not read.
The film's words are written in the language of the page you start from: English here, Chinese on the /zh pages. The README itself can be in either language.
One README, line by line
From a real run on 27 September 2026 with RuoqiJin/mac-auto-bridge. The first column is the repository's own wording; the other two are what LaunchReel AI wrote from it.
| In the README | In the brief | In the storyboard |
|---|---|---|
| “State-aware macOS GUI automation proxy for AI agents.” | One line: “State-aware macOS GUI automation proxy for AI agents.” | Shot 1: big word “Automate”, caption “State-aware GUI automation for AI coding agents.” |
| “Pure Swift. Zero external dependencies.” | Selling point: “Built in pure Swift with zero external dependencies” | Shot 2: tagline “Pure Swift. Zero deps.” |
| “it exposes 20 MCP tools” | Selling point: “20 MCP tools for macOS GUI automation” | Shot 3: dashboard panel, “MCP Tools · 20” |
| “Uses macOS Accessibility API, Vision framework OCR, and CGEvent synthesis.” | Selling point: “Combines the macOS Accessibility API, Vision framework OCR, and CGEvent synthesis” | Shot 4: three glass cards, one per technique |
| “AI coding agents (Claude Code, Cursor, etc.)” | Audience: developers building AI coding agents | Shot 5: big word “Bridge”, caption “For agents like Claude Code and Cursor” |
Storyboard
- 01 · 0:00–0:04Big-word shot: “Automate” with the repository page as the screen.“State-aware GUI automation for AI coding agents.”
- 02 · 0:04–0:07Tagline tracks in on the dark stage.“Pure Swift. Zero deps.”
- 03 · 0:07–0:11Dashboard kit: one KPI panel titled “Automation Toolkit”.“MCP Tools · 20”
- 04 · 0:11–0:15.5Glass-trio kit: three cards, one per technique the README names.“Accessibility API · Vision OCR · CGEvent Synthesis”
- 05 · 0:15.5–0:19.5Big-word shot: “Bridge” over the README's How It Works section.“For agents like Claude Code and Cursor”
- 06 · 0:19.5–0:23Two title lines, the second in the accent colour.“One proxy. Full macOS control.”
- 07 · 0:23–0:30End card: initials “MB”, product name and a line under it.“MacAutoBridge · GUI automation for AI agents”
Written by LaunchReel AI on 27 September 2026 from the public repo RuoqiJin/mac-auto-bridge, an open-source project of ours. Before confirming the brief we removed two things the reader had picked up from GitHub itself (GitHub's logo and a github.com/features page); the storyboard is otherwise unedited. The brief took about a minute and the storyboard about four more. It has not been rendered to a film yet.
Questions
Which Markdown features are read?
Whatever the code host shows on the repository page: headings, paragraphs, lists, code blocks, tables and images, plus up to three linked pages. Badges and images from other sites can appear as screen candidates; untick the ones you don't want on the brief.
What language is the video in?
The language of the page you start from: English on launchreel.ai, Chinese on the /zh pages. The README can be in English or Chinese either way.
Can I get a GIF for the README?
Not today. The film is an MP4 (16:9, 1080p or 4K after export). Most projects post the MP4 with the release, on their docs site or in social posts, and link to it from the README.
Try it on your own project
Paste your link. The brief and the storyboard come back in a few minutes, for you to correct before anything renders. See pricing.