The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
Electron MCP Server
MCP server that lets AI assistants automate, debug and screenshot any Electron app over the Chrome DevTools Protocol — no app changes required.
Install
One-click install
Runs locallyclaude mcp add electron-mcp-server -- npx -y electron-mcp-serverButtons open the app on your computer and ask you to confirm before anything is added.
npx -y electron-mcp-server{
"mcpServers": {
"electron-mcp-server": {
"command": "npx",
"args": [
"-y",
"electron-mcp-server"
]
}
}
}Install command taken from the project's README or registry entry. Always review code before running it.
README excerpt
Read full README on GitHub →Electron MCP Server
Control any Electron app with natural language. An MCP server that lets AI assistants (Claude, Cursor, VS Code, …) drive your Electron app: click buttons, fill forms, take screenshots, read console logs, and run JavaScript inside it — all over the Chrome DevTools Protocol, with no changes to your app.
npx -y electron-mcp-server
Then ask your AI: "Click the Settings button in my Electron app and screenshot the result."
Why you might want it
- 🤖 AI-driven UI automation — describe the interaction ("click Submit", "fill the email field"); no selector wrangling
- 🐛 Debugging without DevTools — console logs and page structure pulled straight into the chat
- 📸 Screenshots on demand — returned as image data the AI can actually look at
- ⚡ Zero app changes — attaches over CDP to any Electron app started with
--remote-debugging-port=9222 - 🔒 Security levels — strict → development, so you choose how much the AI may do
- 🧪 Automated E2E flows — turn manual test steps into repeatable AI-run scenarios
The 4 tools it exposes: get_electron_window_info, take_screenshot, send_command_to_electron (UI interaction + eval), read_electron_logs.
Demo
See the Electron MCP Server in action:
Watch how easy it is to automate Electron applications with AI-powered MCP commands.
🔒 Security & Configuration
Configurable security levels to balance safety with functionality:
Security Levels
- 🔒 STRICT: Maximum security for production environments
- ⚖️ BALANCED: Default security with safe UI interactions (recommended)
- 🔓 PERMISSIVE: More functionality for trusted environments
- 🛠️ DEVELOPMENT: Minimal restrictions for development/testing
Environment Configuration
Configure the security level and other settings through your MCP client configuration:
VS Code MCP Settings:
{
"mcp": {
"servers": {
"electron": {
"command": "npx",
"args": ["-y", "electron-mcp-server"],
"env": {
"SECURITY_LEVEL": "balanced",
"SCREENSHOT_ENCRYPTION_KEY":"your-32-byte-hex-string"
}
}
}
}
}
Claude Desktop Configuration:
{
"mcpServers": {
"electron": {
"command": "npx",
"args": ["-y", "electron-mcp-server"],
"env": {
"SECURITY_LEVEL": "balanced",
"SCREENSHOT_ENCRYPTION_KEY":"your-32-byte-hex-string"
}
}
}
}
Alternative: Local .env file (for development):
# Create .env file in your project directory
SECURITY_LEVEL=balanced
SCREENSHOT_ENCRYPTION_KEY=your-32-byte-hex-string
Security Level Behaviors:
| Level | UI Interactions | DOM Queries | Property Access | Assignments | Function Calls | Risk Threshold |
|---|---|---|---|---|---|---|
strict |
❌ Blocked | ❌ Blocked | ✅ Allowed | ❌ Blocked | ❌ None allowed | Low |
balanced |
✅ Allowed | ✅ Allowed | ✅ Allowed | ❌ Blocked | ✅ Safe UI functions | Medium |
permissive |
✅ Allowed | ✅ Allowed | ✅ Allowed | ✅ Allowed | ✅ Extended UI functions | High |
development |
✅ Allowed | ✅ Allowed | ✅ Allowed | ✅ Allowed | ✅ All functions | Critical |
Environment Setup:
Similar plugs
Reverse engineer anything with agents, from app behavior down to native binaries.
Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype…
Open-source Python library for AI browser agents, plus Browser Use Cloud: stealth browsers and a hosted agent API.