obs-showrunner-mcp
MCP server enabling LLMs to control OBS Studio through high-level show and effect APIs.
README
OBS ShowRunner MCP Server
AI Director in the Loop - MCP server enabling LLMs to control OBS Studio through high-level "show" and "effect" APIs.
Overview
OBS ShowRunner transforms your LLM (Claude, ChatGPT, etc.) into an AI Director that can:
- 🎬 Control show flow - Start/end shows, switch segments with pre-configured scenes
- 🎨 Trigger effects - Visual effects, overlays, and celebratory animations
- 🎵 Manage audio - Switch between mood-based audio profiles (talk, hype, cinema)
- 📷 See the stream - Take screenshots for visual decision-making
- ✏️ Update content - Dynamically change text, browser sources, and images
- 🔒 Safe by default - Dangerous operations are blocked in strict mode
Quick Start
Prerequisites
- OBS Studio 31+ with WebSocket enabled (default port: 4455)
- Node.js 18+
Installation
# Install globally via npm
npm install -g obs-showrunner-mcp
# Or use npx directly (no installation required)
npx obs-showrunner-mcp
Configure Claude Desktop
Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"obs-showrunner": {
"command": "npx",
"args": ["-y", "obs-showrunner-mcp"],
"env": {
"OBS_WEBSOCKET_URL": "ws://localhost:4455",
"OBS_WEBSOCKET_PASSWORD": "your_password"
}
}
}
}
Or if installed globally:
{
"mcpServers": {
"obs-showrunner": {
"command": "obs-showrunner-mcp",
"env": {
"OBS_WEBSOCKET_URL": "ws://localhost:4455",
"OBS_WEBSOCKET_PASSWORD": "your_password"
}
}
}
}
Usage
Once configured, you can ask Claude things like:
- "Start the show and switch to the gaming segment"
- "Take a screenshot of the current stream"
- "Switch audio to hype mode"
- "Show the title overlay with text 'Welcome!'"
- "Mark this moment as a highlight"
Available Tools
Scene Control
| Tool | Description |
|---|---|
get_scene_list |
Get list of available scenes |
set_scene |
Switch to a specific scene |
Show Control
| Tool | Description |
|---|---|
start_show |
Start a show from a template |
end_show |
End the current show |
switch_segment |
Switch to a different segment |
extend_segment |
Extend current segment timer |
get_current_show_state |
Get current show state |
Audio & Effects
| Tool | Description |
|---|---|
set_audio_mood |
Apply audio mood profile (talk, hype, cinema, etc.) |
trigger_effect |
Trigger visual effects |
show_overlay |
Show an overlay |
hide_overlay |
Hide an overlay |
mark_highlight |
Mark a highlight timestamp |
Vision & Content
| Tool | Description |
|---|---|
take_stream_snapshot |
Capture screenshot (Vision) |
update_source_content |
Update text/browser/image sources |
Admin
| Tool | Description |
|---|---|
get_obs_health |
Check OBS connection status |
reconnect_obs |
Reconnect to OBS WebSocket |
set_safety_mode |
Change safety mode |
get_debug_config |
Get debug configuration |
Resources
| URI | Description |
|---|---|
obs://state/current |
Current show state (JSON) |
Configuration
Environment variables:
| Variable | Default | Description |
|---|---|---|
OBS_WEBSOCKET_URL |
ws://localhost:4455 |
OBS WebSocket URL |
OBS_WEBSOCKET_PASSWORD |
- | OBS WebSocket password |
SAFETY_MODE |
strict |
Safety mode (strict/normal/debug) |
ALLOW_STOP_STREAMING |
false |
Allow stopping stream |
ALLOW_STOP_RECORDING |
false |
Allow stopping recording |
OBS_MIC_INPUT_NAME |
Mic/Aux |
Microphone source name |
OBS_BGM_INPUT_NAME |
BGM |
BGM source name |
OBS_GAME_INPUT_NAME |
Game Audio |
Game audio source name |
OBS_SE_INPUT_NAME |
Sound Effects |
Sound effects source name |
Development
# Clone the repository
git clone https://github.com/takurot/obs-showrunner-mcp.git
cd obs-showrunner-mcp
# Install dependencies
npm install
# Run tests
npm test
# Run tests with coverage
npm run test:coverage
# Development mode
npm run dev
# Build
npm run build
Safety Modes
- strict (default): Blocks all dangerous operations
- normal: Allows configured operations only
- debug: Dry-run mode, operations are logged but not executed
Architecture
┌─────────────┐ ┌──────────────────────┐ ┌─────────────┐
│ Claude │────▶│ MCP Server │────▶│ OBS Studio │
│ Desktop │◀────│ (obs-showrunner-mcp) │◀────│ WebSocket │
└─────────────┘ └──────────────────────┘ └─────────────┘
│
┌─────────┴─────────┐
│ │
┌─────▼─────┐ ┌───────▼───────┐
│ Show │ │ Safety │
│ State │ │ Guard │
└───────────┘ └───────────────┘
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。