Talk to Devin

A voice-based planning interface for Devin that uses DeepWiki for codebase context.

Quick Start

1. Copy .env.example to .env and add your API keys:

   cp .env.example .env

   Required keys:

   • OPENAI_API_KEY - For Whisper speech-to-text
   • ANTHROPIC_API_KEY - For Claude AI
   • DEVIN_API_KEY - Your Devin API key

2. Install dependencies:

   npm install

3. Start the app:

   npm run dev

4. Open http://localhost:5173 in your browser

How to Use

1. Enter a repository URL (GitHub, GitLab, etc.)
2. Click "Start Planning"
3. Click the microphone button and start talking about what you want to build
4. See your words transcribed live as you speak
5. AI responds with questions and generates tasks for Devin

Features

• Live speech transcription as you talk
• Conversational AI planning with Claude
• DeepWiki integration for codebase context (coming soon)
• Task generation for Devin

Architecture

• Frontend: Vite + React + TypeScript (port 5173)
• Backend: Express + TypeScript (port 3001)
• Voice: OpenAI Whisper (STT) + Claude + OpenAI TTS
• Context: DeepWiki MCP for codebase knowledge

Next Steps

To complete the integration:

1. DeepWiki MCP Integration - Connect to Cognition's DeepWiki MCP server in backend/src/services/mcp-client.ts
2. Claude MCP Support - Wire up MCP tools to Claude's API calls
3. Task Generation - Add structured task output format
4. Devin API - Submit generated tasks to Devin