yarinkaul/mcp-rubberducking-server

MCP Rubberducking Server

An MCP server that provides AI-powered "rubberducking" service using Gemini 2.5 Pro and GPT-5 for alternative perspectives on complex problems.

Installation

npm install

Configuration

1. Copy .env.example to .env:

cp .env.example .env

2. Add your API keys to .env:

• OPENAI_API_KEY: Your OpenAI API key for GPT-5 access
• GOOGLE_API_KEY: Your Google AI API key for Gemini access

You can use either one or both models.

Building

npm run build

Usage with Claude Code

Important: This server must be run locally, not installed from npm.

1. First, build the server:

npm run build

2. Add to your Claude Code settings (.claude/claude_desktop_config.json):

{
  "mcpServers": {
    "rubberducking": {
      "command": "node",
      "args": ["/Users/yarin/work/rubberduck/dist/index.js"],
      "env": {
        "OPENAI_API_KEY": "your-key",
        "GOOGLE_API_KEY": "your-key"
      }
    }
  }
}

Note: Replace /Users/yarin/work/rubberduck/dist/index.js with the actual absolute path to your built server.

Tool Usage

The server provides one tool: get_second_opinion

This tool will be automatically used by Claude Code when:

• Stuck on a complex problem
• Running in circles without progress
• Facing ambiguous requirements
• Encountering unexpected behavior
• Needing validation on approach
• Debugging complex issues

Parameters

• problem_description (required): Clear description of the problem
• context (required): Relevant code, errors, or technical context
• attempted_solutions (optional): What has been tried
• specific_questions (optional): Specific questions to focus on
• use_models (optional): Which models to use - 'gemini', 'gpt5', or 'both' (default: 'both')

Development

npm run dev

License

MIT