Agentboard

v1.0.0Coding Agentsstable

A switchboard for AI in your browser: wire in any model, script WebMCP tools, connect remote MCP servers, bring your commands.

agentboardmcpai-integration
Share:
115
Stars
0
Downloads
0
Weekly
0/5

What is Agentboard?

Agentboard is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to switchboard for ai in your browser: wire in any model, script webmcp tools, connect remote mcp servers, bring your commands.

A switchboard for AI in your browser: wire in any model, script WebMCP tools, connect remote MCP servers, bring your commands.

This server falls under the Coding Agents category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • A switchboard for AI in your browser: wire in any model, scr

Use Cases

Wire together multiple AI models in your browser for agent workflows.
Connect remote MCP servers and script WebMCP tools in one interface.
Build custom command sets and agent orchestration dashboards.
igrigorik

Maintainer

LicenseMIT
Languagetypescript
Versionv1.0.0
UpdatedMay 4, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx agentboard

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use Agentboard

AgentBoard is a Chrome browser extension that acts as a switchboard for AI models and MCP tools, letting you wire together multiple AI providers (OpenAI, Anthropic, Google, Ollama, and any OpenAI-compatible endpoint) with browser-native WebMCP tools and remote MCP servers — all without a backend. It ships with built-in tools for reading the current page as markdown, extracting YouTube transcripts, and fetching external URLs, and allows developers to author custom WebMCP tool scripts that interact with the DOM. Power users and developers who want a single browser-based interface to orchestrate agents across many models and tools will find it a flexible alternative to desktop MCP clients.

Prerequisites

  • Google Chrome browser (or Chromium-based browser)
  • API key for at least one supported AI provider: OpenAI, Anthropic (Claude), Google Gemini, or a local Ollama instance
  • Node.js 18+ and pnpm if building from source for development
  • Remote MCP servers must expose an HTTP streaming endpoint if you want to connect them to AgentBoard
1

Install AgentBoard from the Chrome Web Store

Search for 'AgentBoard' in the Chrome Web Store and click Install, or use the direct link from the project's GitHub page. The extension is added to Chrome and accessible from the toolbar.

2

Open AgentBoard and configure an AI provider

Click the AgentBoard icon in the Chrome toolbar. Go to Settings and add your API key for your preferred provider. You can add multiple providers and switch between them per agent profile.

3

Create an agent profile

Set up an agent profile with a system prompt, model selection, temperature, and optional reasoning settings. Profiles let you have specialized agents (e.g., a coding assistant vs. a summarizer) with their own configurations.

4

Enable built-in WebMCP tools

Toggle on the built-in tools: agentboard_read_page (converts the current page to markdown), agentboard_youtube_transcript (extracts video transcripts), and agentboard_fetch_url (fetches external URLs). These work on any page without additional setup.

5

Connect a remote MCP server (optional)

If you have a remote MCP server, add its HTTP streaming URL and optional bearer token in the MCP Servers settings panel. AgentBoard will discover and expose all tools from that server to your agents.

6

Write a custom WebMCP tool script (optional)

Author a custom tool by writing a JavaScript snippet that interacts with the DOM. Scope it to specific URL patterns (e.g., github.com/*) so it activates only on relevant pages.

// Example: extract all links from the current page
export const tool = {
  name: 'get_page_links',
  description: 'Returns all links on the current page',
  urlPattern: '*://*/*',
  execute: () => {
    return Array.from(document.querySelectorAll('a'))
      .map(a => ({ text: a.innerText, href: a.href }));
  }
};

Agentboard Examples

Client configuration

Example agent profile configuration for connecting to an Ollama local model in AgentBoard's settings JSON format.

{
  "provider": "openai",
  "endpoint": "http://localhost:11434/v1",
  "model": "llama3.1:70b",
  "apiKey": "ollama",
  "temperature": 0.7,
  "systemPrompt": "You are a helpful assistant with access to browser tools."
}

Prompts to try

Example commands and prompts to use inside AgentBoard with the built-in tools enabled.

- "Read this page and summarize the main points"
- "/tldr" (built-in command to bullet-point summarize the current page)
- "Get the transcript from this YouTube video and identify key timestamps"
- "Fetch the content from https://example.com/api/data and parse the JSON"
- "/explain how async/await works" (expands to a teaching-style prompt)
- "Extract all the links on this page and categorize them by domain"

Troubleshooting Agentboard

Extension does not appear in the Chrome toolbar after installation

Click the puzzle piece icon in the Chrome toolbar to see all extensions, then pin AgentBoard. If the extension is not listed, go to chrome://extensions and verify it is enabled.

AI provider returns authentication error

Double-check the API key in AgentBoard Settings matches your provider's dashboard. For Anthropic, ensure the key starts with 'sk-ant-'. For Ollama, the API key can be set to any non-empty string like 'ollama' and the endpoint should be http://localhost:11434/v1.

Remote MCP server tools do not appear after adding the server URL

Verify the remote MCP server supports HTTP streaming (SSE) transport. Check that the bearer token is correct if required. Open the browser's DevTools Network tab while AgentBoard loads to inspect any connection errors to the MCP endpoint.

Frequently Asked Questions about Agentboard

What is Agentboard?

Agentboard is a Model Context Protocol (MCP) server that switchboard for ai in your browser: wire in any model, script webmcp tools, connect remote mcp servers, bring your commands. It connects AI assistants to external tools and data sources through a standardized interface.

How do I install Agentboard?

Follow the installation instructions on the Agentboard GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with Agentboard?

Agentboard works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is Agentboard free to use?

Yes, Agentboard is open source and available under the MIT license. You can use it freely in both personal and commercial projects.

Browse More Coding Agents MCP Servers

Explore all coding agents servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "agentboard": { "command": "npx", "args": ["-y", "agentboard"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use Agentboard?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides