arc-mcp
Enables semantic browser automation for the Arc browser, allowing users to perform web tasks like searching, shopping, and data extraction through natural language commands.
README
<div align="center">
Playwright-style semantic browser automation for Arc
<p> <img src=".github/assets/claude-logo.svg" alt="Claude" height="20" /> </p>
</div>
Installation
Claude Desktop
- Download the latest
.mcpbfile from releases - Open Claude Desktop
- Go to Settings → Extensions
- Drag and drop the
.mcpbfile onto the window
Claude Code
claude mcp add --scope user --transport stdio arc nix run github:andrewgazelka/arc-mcp
Or manually add to your MCP settings:
{
"mcpServers": {
"arc": {
"command": "nix",
"args": ["run", "github:andrewgazelka/arc-mcp"]
}
}
}
Use Cases
<img src=".github/assets/icon-flight.svg" width="16" height="16" /> Flight Search — Compare prices across airlines, check availability, book tickets
<img src=".github/assets/icon-shopping.svg" width="16" height="16" /> Shopping — Search products, read reviews, add to cart, checkout
<img src=".github/assets/icon-groceries.svg" width="16" height="16" /> Groceries — Navigate store catalog, build cart, schedule delivery
<img src=".github/assets/icon-research.svg" width="16" height="16" /> Research — Open multiple sources, extract data, synthesize information
Features
Node.js REPL
Execute automation scripts with full access to the browser API — no predefined tool constraints.
await browser.click({ role: { role: 'button', name: 'Search' } });
await browser.fill({ label: 'Email' }, 'user@example.com');
const tabs = await browser.listTabs();
return tabs.length;
Semantic Locators
Find elements the way users see them — by role, label, text, or placeholder. No more fragile CSS selectors.
await browser.click({ role: { role: 'button', name: 'Search' } });
await browser.fill({ label: 'Email' }, 'user@example.com');
await browser.type({ placeholder: 'Search...' }, 'London');
Smart DOM Tree
Get a structured, semantic representation of the page with only interactive elements.
const structure = await browser.getPageStructure(10);
Available Functions
The browser object provides:
Semantic Actions:
click(locator, options?)— Click element by role, label, text, placeholder, or CSSfill(locator, value, options?)— Fast fill (replaces entire value)type(locator, text, options?)— Type character-by-character (triggers autocomplete)selectOption(locator, option, options?)— Select dropdown option
Navigation:
openUrl(url, newTab?)— Open URL in new or current tabgetCurrentTab()— Get active tab info (title, URL)listTabs()— List all open tabsswitchToTab(tabId)— Switch to tab by indexcloseTab(tabId)— Close tab by indexreloadTab(tabId?)— Reload tabgoBack(tabId?)— Navigate backgoForward(tabId?)— Navigate forward
Page Analysis:
getPageStructure(maxDepth?, tabId?)— Get semantic DOM tree
Low-Level:
executeJavaScript(code, tabId?)— Run arbitrary JavaScript
Example Workflow
// Open Google Flights
await browser.openUrl('https://www.google.com/travel/flights');
// Fill origin
await browser.fill({ label: 'Where from?' }, 'San Francisco');
// Type destination (triggers autocomplete)
await browser.type({ placeholder: 'Where to?' }, 'London');
// Click search
await browser.click({ role: { role: 'button', name: 'Search' } });
Architecture
This is a monorepo with three packages:
- @arc-mcp/applescript — Low-level AppleScript bridge for Arc
- @arc-mcp/browser — High-level Node.js API with semantic locators
- arc-mcp-server — MCP server exposing REPL
See ARCHITECTURE.md for detailed design.
Requirements
- macOS
- Arc browser
Development
npm install
npm run build # Build all packages
npm run typecheck # Type check all packages
npm run pack # Create .mcpb bundle
Package Usage
Each package can be used independently:
# Use browser API in your own scripts
npm install @arc-mcp/browser
# Use AppleScript bridge directly
npm install @arc-mcp/applescript
<div align="center">
Not affiliated with The Browser Company. Arc is a trademark of The Browser Company.
MIT · Andrew Gazelka
</div>
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.