Happy slice of mcp pizzaMCP.pizza

kibitz

MCP.Pizza Chef: nick1udwig

Kibitz is a client MCP entity designed as a professional coding agent that integrates with various MCP servers to provide developers with enhanced coding workflows. It supports configuration per project, connects to MCP servers via WebSocket, and requires API keys for LLM providers like Anthropic. Kibitz streamlines coding tasks by leveraging real-time context and model interactions within a developer-friendly UI.

Use This MCP client To

Connect to MCP servers for real-time coding context Configure project-specific LLM API keys and system prompts Manage multiple MCP server connections via WebSocket Run a development server with a web UI for coding assistance Use Anthropic API integration for advanced language model features Clone and set up the client for local development and customization

README

kibitz

The coding agent for professionals

et6EDGkZeo4ZQFeX.mp4

Prerequisites

  • git
  • npm

Installation

  1. Clone the repository:
git clone https://github.com/nick1udwig/kibitz.git
cd kibitz
  1. Install dependencies:
npm install
  1. Run the development server:
npm run dev
  1. Open http://localhost:3000 in your browser.

Configuration

  1. Open the Settings panel in the UI
  2. Enter your Anthropic API key (Get one here).
  3. Optionally set a system prompt
  4. Configure MCPs by running them using ws-mcp and then connecting to them in the Settings page

Note configuration is PER-PROJECT. When creating a new project, it will use some, but not all, of the current project's configuration: the API key, model, and system prompt will be copied over, but MCP servers will not.

Building for Kinode

  1. Add a base to the endpoint by building with the NEXT_PUBLIC_BASE_PATH (MUST start with a /),
  2. Change the default WS-MCP server URI by specifying NEXT_PUBLIC_DEFAULT_WS_URI (MUST start with a /),

like so:

NEXT_PUBLIC_BASE_PATH=/kibitz:kibitz:nick.kino NEXT_PUBLIC_DEFAULT_WS_URI=/fwd-ws:kibitz:nick.kino npm run build

and then copy the contents of out/ into the package's pkg/ui/ dir.

kibitz FAQ

How do I install kibitz?
Clone the repository, install dependencies with npm, then run the development server using 'npm run dev'.
How do I configure API keys in kibitz?
Open the Settings panel in the UI and enter your Anthropic API key; this is required for LLM access.
Can kibitz connect to multiple MCP servers?
Yes, you can configure and connect to multiple MCP servers via WebSocket in the Settings page.
Is the configuration shared across projects?
No, configuration is per-project; API keys and prompts copy over, but MCP server connections do not.
What prerequisites are needed to run kibitz?
You need git and npm installed to clone and run the client.
Does kibitz support LLM providers other than Anthropic?
While Anthropic is the primary example, kibitz can be configured to work with other LLM providers like OpenAI and Claude through MCP servers.
How do I start the kibitz development server?
After installing dependencies, run 'npm run dev' and open http://localhost:3000 in your browser.
Can I customize the system prompt in kibitz?
Yes, you can set a custom system prompt per project in the Settings panel.