Kibitz MCP Server

Kibitz is a specialized coding assistant designed for professional developers. It provides an intuitive interface for interacting with large language models like Claude from Anthropic, with a focus on enhancing coding workflows. What sets Kibitz apart is its support for the Model Context Protocol (MCP), allowing it to connect to external tools and services that extend the AI's capabilities. This enables developers to leverage AI assistance while maintaining access to their local development environment and tools.

Overview

Kibitz is a web-based coding assistant that connects to Anthropic's Claude models and supports the Model Context Protocol (MCP) for enhanced functionality. It's designed to help professional developers with coding tasks while providing a clean, intuitive interface.

Installation

There are two main ways to use Kibitz:

Option 1: Use the hosted version

Visit kibi.tz to use the hosted version of Kibitz.

Option 2: Run locally

To run Kibitz locally:

Ensure you have Git and npm installed on your system

Clone the repository:

git clone https://github.com/nick1udwig/kibitz.git
cd kibitz

Install dependencies:
```
npm install
```
Start the development server:
```
npm run dev
```
Open http://localhost:3000 in your browser

Configuration

Kibitz requires configuration before use:

Open the Settings panel in the Kibitz UI
Enter your Anthropic API key (you can get one here)
Optionally set a system prompt to customize the assistant's behavior
Configure MCP servers if you want to extend functionality (see MCP Configuration below)

Note that configuration is per-project. When creating a new project, some settings (API key, model, system prompt) will be copied over, but MCP server connections will not.

MCP Configuration

Kibitz supports the Model Context Protocol (MCP), which allows the AI to interact with external tools and services. To use MCP with Kibitz:

Run MCP servers using ws-mcp
In the Kibitz Settings panel, connect to your running MCP servers
The AI can now use the tools provided by those servers during your conversations

Building for Kinode

If you want to build Kibitz for use with Kinode:

Build with the appropriate environment variables:
```
NEXT_PUBLIC_BASE_PATH=/kibitz:kibitz:nick.kino NEXT_PUBLIC_DEFAULT_WS_URI=/fwd-ws:kibitz:nick.kino npm run build
```
Note: NEXT_PUBLIC_BASE_PATH must start with a / and NEXT_PUBLIC_DEFAULT_WS_URI must also start with a /
Copy the contents of the out/ directory into the package's pkg/ui/ directory

Usage Tips

Kibitz saves your configuration and conversation history per project
You can create multiple projects for different coding tasks
The interface supports markdown, code highlighting, and file uploads
Use the MCP connections to give the AI access to your development environment

Kibitz MCP Server

ws-mcp