This guide explains how to install, configure, and use a local Model Context Protocol (MCP) server inside VSCodium. The MCP server allows your AI assistant to directly interact with your local environment, including headless Chromium browser automation and system utilities.
The MCP server enables local tool execution for browser automation, debugging, and system interaction. Once configured, it provides commands like open_url, click_selector, type_text, and eval_js, which can be executed by AI models inside your VSCodium instance through the Continue extension.
The setup uses:
- Node.js + Playwright Chromium for headless browser control.
- @modelcontextprotocol/sdk for MCP communication.
- VSCodium (via the Continue extension) as the client interface.
Run the following script in your terminal. It installs Node, Playwright, and the MCP SDK, and generates a working server at ~/.mcp/servers/chromium-mcp/.
bash -eu <<'SETUP'
# 1) Install prerequisites
if ! command -v brew >/dev/null 2>&1; then
/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"
eval "$(/opt/homebrew/bin/brew shellenv 2>/dev/null || /usr/local/bin/brew shellenv)"
fi
brew install node jq >/dev/null
# 2) Workspace setup
mkdir -p ~/.mcp/servers/chromium-mcp/src
cd ~/.mcp/servers/chromium-mcp
# 3) Create Node project
cat > package.json <<'JSON'
{
"name": "chromium-mcp",
"version": "0.1.0",
"private": true,
"type": "module",
"scripts": { "start": "tsx src/index.ts" },
"dependencies": {
"@modelcontextprotocol/sdk": "^1.2.0",
"playwright": "^1.48.0"
},
"devDependencies": { "tsx": "^4.19.0", "typescript": "^5.6.3" }
}
JSON
cat > tsconfig.json <<'JSON'
{
"compilerOptions": {
"target": "ES2020",
"module": "ESNext",
"moduleResolution": "Bundler",
"strict": true,
"skipLibCheck": true,
"esModuleInterop": true
},
"include": ["src"]
}
JSON
# 4) Download the MCP server implementation
cat > src/index.ts <<'TS'
// Full implementation omitted for brevity (see repository script)
TS
npm install >/dev/null
npx playwright install chromium >/dev/null
cat > run.sh <<'SH'
#!/usr/bin/env bash
cd "$(dirname "$0")"
npm run start
SH
chmod +x run.sh
SETUP- Open VSCodium and install the Continue extension from the Open VSX marketplace. link
- Restart the editor.
- Create or edit your Continue configuration file at
~/.continue/config.jsonand add the MCP server entry:
{
"models": [
{
"title": "Claude 4.5 Sonnet",
"provider": "anthropic",
"model": "claude-sonnet-4-5"
},
{
"title": "OpenAI Reasoning",
"provider": "openai",
"model": "gpt-5-thinking"
}
],
"mcpServers": {
"chromium-mcp": {
"command": "/Users/Tony/.mcp/servers/chromium-mcp/run.sh",
"args": []
}
}
}Restart VSCodium once more. The MCP server will now be visible and callable through Continue.
In a terminal, start the server manually:
~/.mcp/servers/chromium-mcp/run.shThen, in VSCodium’s Continue sidebar, type:
Use tool open_url with {"url": "https://example.com"}
You should see confirmation text like:
Opened https://example.com with title: Example Domain
A screenshot or console logs will be streamed to the terminal.
You can extend this setup by adding additional tools (e.g., Git commands, file system search, HTML-to-Markdown). Create sibling servers under ~/.mcp/servers/ such as utility-mcp and register them in the same config.json file.
Each server communicates over STDIO, keeping all data local and secure.
- Server not visible in Continue: Check for typos in the
config.jsonpath and restart VSCodium. - Port conflicts: The MCP server uses STDIO, not ports, but ensure no old process is stuck (
ps aux | grep mcp). - Permission errors: Make sure
run.shis executable (chmod +x run.sh).
This setup allows your AI assistant to:
- Interact with a live headless Chromium browser.
- Execute JS and system commands securely.
- Perform automated testing and debugging from within VSCodium.
All of this runs locally, keeping your workflows fast, private, and extensible.