A VS Code extension that implements an intelligent chat participant @dspy inside Copilot Chat, using DSPy-style reasoning loops to provide more reliable and structured code assistance.

Yu enhances your development workflow by adding a @dspy chat participant that uses structured reasoning loops (plan → draft → critique → revise) to provide more reliable code completions and refactoring suggestions. The extension leverages Ax (TypeScript DSPy) to orchestrate multi-step LLM conversations through the VS Code Language Model API.

• Structured Reasoning: Uses DSPy-style patterns to reduce hallucinations and improve code quality
• Context-Aware: Automatically collects relevant code context from your editor
• Streaming Responses: Real-time markdown output with "Insert to Editor" actions
• Extensible: Built to be composable and adaptable to new AI providers

• 🤖 @dspy Chat Participant: Invoke from Copilot Chat for structured assistance
• 📋 Multi-Step Process: Responses include Plan, Draft, Critique, and Final sections
• 🎯 Smart Context: Automatically uses selected code or current file as context
• ⚡ Quick Actions: "Insert to Editor" button to apply final code snippets
• 🔄 Ax Pipeline: Uses @ax-llm/ax for structured LLM orchestration

• 💭 Conversation Memory: Remember findings and instructions across chat turns
• 🔍 Extended Context: Integration with docs and repository context
• 🚀 Proactive Suggestions: Intelligent next-step recommendations
• 🏠 Local Runners: Support for local DSPy (Python) execution
• See roadmap for more details.

• VS Code 1.103.0 or higher
• Access to VS Code's Language Model API (varies by organization/subscription)

1. Clone and install dependencies:

   npm install

2. Compile and package the extension:

   npm run package

3. Install the extension in VS Code:

   • Open VS Code
   • Press Ctrl+Shift+P (Windows/Linux) or Cmd+Shift+P (macOS)
   • Type "Extensions: Install from VSIX..." and select the generated .vsix file

1. Open Copilot Chat in VS Code

2. Type @dspy followed by your request, for example:

   @dspy Refactor this function to be more efficient
   @dspy Add error handling to this code
   @dspy Explain what this code does and suggest improvements
   

3. The @dspy participant will respond with structured reasoning steps

4. Use the "Insert to Editor" button to apply the final code suggestions

src/
├── extension.ts          # Main extension entry point
└── test/
    └── extension.test.ts # Extension tests
docs/
├── prd.md               # Product Requirements Document
└── todos.md            # Development TODO list

# Development
npm run compile          # Compile TypeScript, check types, lint, and build
npm run watch            # Run development mode with file watching
npm run package          # Create production build for publishing

# Quality Assurance
npm run check-types      # Run TypeScript type checking
npm run lint             # Run ESLint
npm run test             # Run VS Code extension tests

# Watch Commands
npm run watch:esbuild    # Watch for changes and rebuild with esbuild
npm run watch:tsc        # Watch for TypeScript changes
npm run watch-tests      # Watch for test file changes

• Ensure tasks. and launch.json are configured in .vscode
• Build: npm run compile
• Run: “Run Extension (Watch)” config
• In the Dev Host: open Copilot Chat → @dspy hello
• See streamed ack + “dspy” Output channel logs

# Ensure Claude Code is running
ccr status          # Check if Claude Code Router is running
ccr start           # Run development mode with file watching

# Launching Claude Code (in separate terminal tab)
ANTHROPIC_BASE_URL=http://localhost:3456 ANTHROPIC_API_KEY=any-string-is-ok npx @anthropic-ai/claude-code        

# Taskmaster 
source .env                                       # Ensure API keys are loaded and available as env variables
tm list                                           # List tasks
task-master expand --id=3                         # Expand task 3
task-master set-status --id=3 --status=done       # Expand task 3

The extension follows a modular architecture:

1. Extension Layer: Registers the chat participant and manages context collection
2. Ax Pipeline: Uses structured signatures for plan/draft/critique orchestration
3. VS Code LM Adapter: Bridges Ax with the VS Code Language Model API
4. UI Integration: Handles streaming responses and quick actions

• Target: ES2022
• Module: Node16
• Strict mode enabled
• Source maps enabled

• TypeScript naming conventions
• Code quality enforcement (curly braces, equality checks, semicolons)

• VS Code 1.103.0+
• Access to VS Code Language Model API
• Node.js for development
• No Python runtime required (Ax runs in TypeScript/JavaScript)

Phase: Early Development - MVP in Progress

The project scaffold is complete with:

• ✅ VS Code extension structure
• ✅ Basic command registration
• ✅ TypeScript configuration
• ✅ Build and test pipeline
• ✅ Task Master integration
• ⏳ Chat participant implementation
• ⏳ Ax pipeline integration
• ⏳ VS Code LM API integration

See docs/todos.md for detailed progress tracking.

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests for new functionality
5. Run npm run lint and npm run test
6. Submit a pull request

This project is licensed under the MIT License - see the LICENSE file for details.

• VS Code Extension API
• Ax (TypeScript DSPy)
• DSPy Documentation
• VS Code LM API

Built with ❤️ using Claude Code