# showtime [Health: Active]

**Category:** 💻 Developer Tools  
**Repository:** https://github.com/FavioVazquez/showtime  
**GitHub Stars:** 254  
**npm Downloads (last month):** 43  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/showtime

## Description
Local video studio for your coding agent: scenes, voice, music and captions, on your machine.

## Claude Desktop Quick Installation
Heuristic fallback — verify the package name and runner against the repository README before running it. Uses `npx` (confidence: low):

```json
"mcpServers": {
  "showtime": {
    "command": "npx",
    "args": ["-y","showtime"]
  }
}
```

## Documentation

## What showtime MCP server does

The showtime MCP server adds a local video-production workflow to coding agents. An agent can plan a video, choose or infer production details, create the required project files, and ask the local machine to render the result. Supported output work includes motion graphics, voice-over, music, sound design, captions, footage editing, and platform-specific exports.

The workflow is intended for requests such as making a short launch video for a repository, creating a narrated explanation of a system, or turning a software test failure into a visual report. The project also includes specialized workflows for diagrams, session recaps, and interactive HTML exports.

## How it works

A request starts in the coding agent. In the default quick mode, the agent states its assumptions and begins; studio mode can show a concept, visual treatment, music bed, and storyboard before production. The agent can provide stills or a fast draft before the full-quality render. Quality mode runs a critic review after rendering, while lean mode skips that review unless the video is intended for publication.

Rendering and supporting work happen locally. The README states that the workflow uses no cloud AI services, requires no API keys, and uploads no project material. First-time setup downloads the required components, and larger assets such as the transcription model, Manim, and music tracks are fetched when a video first needs them.

Each request receives a separate timestamped directory under `showtime-out/`. The result can include `final.mp4`, `poster.jpg`, `share.txt`, an `exports/` directory, and a `work/` directory. Files are not overwritten between requests.

## Setup and configuration

Install `uv` and Node.js 24 or 22 LTS; Node.js 20 or newer is supported. Restart the coding agent after installing the prerequisites. Claude Code can add the project with:

```text
/plugin marketplace add FavioVazquez/showtime
/plugin install showtime@showtime
```

For agents that support Agent Skills, the README gives this installation command:

```text
npx skills add FavioVazquez/showtime
```

The showtime MCP server performs setup after the agent confirms the download. The initial installation typically downloads about 0.5–0.75 GB and uses about 1.4–1.8 GB in `~/.showtime`, depending on the system. The first setup usually takes around two to six minutes. A sandbox that blocks downloads may require running setup from a terminal.

## Tools and capabilities

Available workflows and commands described by the project include:

- Create and render narrated videos with voice, music, captions, and edited footage.
- Run `showtime qa` to check a finished video.
- Use `showtime new diagram` for animated system explanations.
- Export a scene as an offline interactive HTML page with `showtime export html <project> --interactive`.
- Use `showtime recap --latest` to create a 30–60 second narrated film from an agent session.
- Use `showtime repro` to turn a Playwright trace into a narrated failure video and Markdown issue summary.
- Produce previews, posters, final MP4 files, and platform exports.

## Limitations and notes

Rendering speed depends on the machine. The project describes a showreel rendered on a 64-core CPU without a GPU, but that result is not a general performance guarantee. Installation and later workflows may download substantial assets and require additional disk space.

The README identifies the project as early 0.4-series software. Tested platforms include Intel Mac, Apple Silicon, Linux x64, Linux arm64, Windows x64, and Windows 11 on Arm; Windows 10/11 desktop support is listed as still needing additional testing. The provided material does not describe cloud rendering, API-key configuration, or a hosted service.

_Full upstream README: https://allmcps.com/mcp/showtime/readme_