AI Decision System
Replayable, offline-verifiable records for AI-assisted go/no-go decisions: experiment promotions, typed model answers and human-approved actions.
코딩 에이전트가 계획·목업·구축을 실시간으로 나란히 보여주며, 무료 오픈소스 Mac 앱 또는 팀 호스팅 제공
Coding agents plan, mock up and build from your own repo, with every version live side by side. Free and open source: a Mac app, or host it for your team.
개발 도구TypeScript · JavaScriptGitHub @zschiller온라인운영자 미확인
써 보셨어요? 한 번만 눌러 주세요. 로그인은 필요 없어요.
README 전체를 불러오는 중입니다. 아래는 저장된 발췌입니다.
From idea to code, on one canvas.
Coding agents plan, mock up and build from your own repo, with every version
live side by side.
The plan sits in a doc, the mockup in a design tool that has never seen your
components, and the build in a terminal and a browser tab. Screenplay puts all
three on one canvas, next to your code.
Type a prompt anywhere on the canvas and an agent makes it right there. It
reads your code first, so whatever it makes fits your app.README 전문 보기 ↗The plan sits in a doc, the mockup in a design tool that has never seen your components, and the build in a terminal and a browser tab. Screenplay puts all three on one canvas, next to your code.
Type a prompt anywhere on the canvas and an agent makes it right there. It reads your code first, so whatever it makes fits your app.
Chats share memory, files and skills, so what you tell one chat, every chat knows.
| Desktop app | Hosted app (you deploy it) | |
|---|---|---|
| For | One person on a Mac with Apple silicon | A team, on infrastructure you deploy |
| Chats run in | Git worktrees on your machine | A sandbox VM per chat |
| Agent | Antigravity, Claude Code, Codex or OpenCode, with the login and subscription you already have | The built-in agent, on Anthropic, OpenAI, Google, the Vercel AI Gateway or any OpenAI-compatible endpoint |
| Together | Single user | Shared canvases, live cursors, comments and live frames |
| Start | Quickstart | Self-hosting guide |
Both are the same product, built from this repository.
You need Node.js 20 or newer, pnpm 9 and git. The local build runs the full product with no external services (embedded Postgres, a local Yjs server, and each chat’s code as a git worktree), plus a coding CLI such as Claude Code for the agent.
git clone https://github.com/zschiller/screenplay.git
cd screenplay
pnpm install
cd apps/app
NEXT_PUBLIC_SCREENPLAY_PROFILE=desktop NEXT_PUBLIC_YJS_HOST=local NEXT_PUBLIC_BASE_PATH= \
SANDBOX_BACKEND=local SCREENPLAY_DB=pglite BLOB_STORE=local-fs AGENT_ENGINE=external \
ENCRYPTION_KEY=$(openssl rand -hex 32) TERMINAL_AUTH_SECRET=$(openssl rand -hex 32) \
pnpm dev
Then open http://localhost:3000. The Development guide covers the multi-user build, the desktop app, and every contributor command.
| Path | What it is |
|---|---|
apps/app | The product: canvas, agent runtime and API routes |
apps/desktop | The Tauri shell that packages apps/app as the Mac app |
apps/docs | The docs site, screenplay.space/docs |
apps/homepage | The website, screenplay.space |
packages/screenplay-knobs | Declare knobs in your app |
packages/screenplay-state | Share state across viewers of a frame |
packages/ui | Shared shadcn/ui components |
Issues and pull requests are welcome. The docs site is the home for all
documentation, so please add and update pages in
apps/docs/content rather than growing this README. The
screenshots above come from the docs and refresh themselves whenever the app
changes.
MIT. See LICENSE.