explaingit

utopai-research/pai-pro

★ 12JavaScriptAudience · vibe coderComplexity · 3/5LicenseSetup · moderate

TLDR

A local filmmaking studio app you run in a browser and control by chatting with an AI coding agent, describe what you want and the agent writes screenplays, generates images, video, and designs characters on your machine.

Mindmap

mindmap
  root((PAI-Pro))
    Interface
      Visual canvas
      Timeline tab
      Embedded terminal
    Capabilities
      Screenplay writing
      Image generation
      Video generation
      Character design
    Setup
      Docker
      PAI Lite API key
      Local browser
    Use Cases
      Short film creation
      Scene planning
      Custom skills
Click or tap to explore — scroll the page freely

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

Things people build with this

USE CASE 1

Describe a short film concept in plain language and let the AI agent generate the screenplay, character images, and a sequence of video clips

USE CASE 2

Arrange generated clips on the timeline and play back a rough cut of a scene without leaving the browser

USE CASE 3

Write custom SKILL.md files to teach the AI agent new filmmaking workflows specific to your project

USE CASE 4

Use the visual canvas to organize characters, locations, and notes into scenes before asking the agent to generate media

Tech stack

JavaScriptReact FlowDocker

Getting it running

Difficulty · moderate Time to first run · 1h+

Requires a PAI Lite API key for media generation, Docker simplifies dependency setup.

Free for personal and internal business use, any commercial use beyond that requires an enterprise license from Utopai Studios.

In plain English

PAI-Pro is a local filmmaking studio application that you control through a coding agent like Claude Code, Codex, Cursor, or Gemini CLI. The idea is that instead of clicking through menus yourself, you describe what you want to a coding agent running inside the app, and the agent carries out filmmaking tasks such as writing a screenplay, generating image and video clips, or designing characters. The interface has three main areas. A visual canvas built with a tool called React Flow lets you place nodes representing characters, locations, images, videos, and notes, then group them into scenes. A timeline tab lets you arrange your clips in order, drag them around, and play them back in sequence. An embedded terminal in the right panel runs your coding agent directly, so you can type instructions without switching to another window. Everything stays on your local machine. Project files live on disk alongside the generated media, and nothing leaves your computer except the actual requests to generate images, video, or voice, which go through a service called PAI Lite using a single API key. Credit is only consumed when you explicitly ask for media generation, browsing and chatting do not cost anything. Setting it up works either through Docker, which handles all the dependencies, or by cloning the repository and running a setup script directly on your machine. Either way, it runs in a browser at a local address. The skills system, described in plain SKILL.md files, defines what the agent knows how to do in filmmaking terms. You can also write your own custom skills. The license allows personal and internal business use. Commercial use beyond that scope requires an enterprise license from Utopai Studios, which built the project.

Copy-paste prompts

Prompt 1
Using PAI-Pro, help me write a SKILL.md file that teaches the AI agent how to generate a three-act screenplay structure from a one-sentence premise.
Prompt 2
In PAI-Pro I have placed several character nodes on the canvas. Help me write an agent prompt that assigns each character a consistent visual style for image generation.
Prompt 3
Help me set up PAI-Pro using Docker on a Mac and configure the PAI Lite API key so I can start generating video clips from a short film script.
Prompt 4
Write a PAI-Pro agent instruction that takes a screenplay scene and breaks it into a sequence of image prompts, one per shot, ready to send to the media generator.
Prompt 5
Using PAI-Pro's timeline tab, help me write an agent workflow that assembles generated clips in screenplay order and plays them back as a rough cut.
Open on GitHub → Explain another repo

← utopai-research on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.