Agent TODOs

Your AI pair programmer's memory system. Seamlessly integrate persistent task tracking with GitHub Copilot and VS Code's native AI features—giving your coding assistant perfect memory across sessions.

"Finally, an AI assistant that remembers what we planned to do next."

Why Agent TODOs?

VS Code's agent mode is incredibly powerful for code generation and problem-solving. Agent TODOs takes this foundation to the next level by adding advanced workflow management capabilities that transform how agent mode handles complex, multi-session projects.

Agent TODOs upgrades your AI coding experience by adding a persistent memory system that:

• Enhances workflow continuity across sessions, building on VS Code's excellent chat capabilities
• Scales task complexity by adding structured decomposition to complement VS Code's problem-solving power
• Optimizes context efficiency by working alongside VS Code's native AI to manage information flow
• Adds professional tracking with rollback capabilities that extend VS Code's debugging features

Key Features

• 🗺️ Persistent AI Memory: Structured task decomposition that survives session restarts—no more lost context or repeated work
• 📊 Workflow State Management: Three-state tracking (pending → in-progress → completed) with priority levels and rich task descriptions
• 🤖 Native VS Code Integration: Built-in todo_read and todo_write tools that work seamlessly with GitHub Copilot and VS Code's agent mode
• 🧠 Context Window Optimization: Isolates active tasks to prevent token overflow while maintaining full project context
• 🔗 Auto-Sync with Copilot: Automatically inject todos into .github/copilot-instructions.md for enhanced AI context awareness
• 🎯 Smart Task Management: Interactive tree view in VS Code's Explorer with one-click status updates and instant AI execution
• 🛡️ Error Recovery: Granular tracking enables rollback, retry, and debugging of failed development steps
• 🔧 Flexible Integration: Works as VS Code extension or standalone MCP server for advanced integrations

How It Works

1. Install → Extension adds todo tools to GitHub Copilot automatically
2. Plan → Ask Copilot to create and organize your development tasks
3. Track → Visual tree view shows progress; AI remembers everything
4. Execute → Run todos directly in VS Code Chat with full context

🔄 Example Workflow

You: "Create a todo list for implementing user authentication"

GitHub Copilot: "I'll create a structured plan for user authentication. Let me break this down into atomic, trackable tasks..."

📝 Creating todos:
• Set up authentication routes (high priority)
• Configure auth library integration  
• Create login/register endpoints
• Add JWT middleware for protected routes
• Write authentication tests

You: "Start with the first task"

GitHub Copilot: "I'll help you set up the authentication routes. First, let me mark this as in-progress to track our workflow state..."

🔄 Updated todo: "Set up authentication routes" → in_progress

*Creates auth.routes.ts file*
*Implements basic route structure*

✅ Marking "Set up authentication routes" as completed

Now moving to the next task in our workflow...

🔄 Updated todo: "Configure auth library integration" → in_progress

*Installs passport and express-jwt packages*
*Sets up auth configuration*

✅ Completed "Configure auth library integration"

You: "What's left to do?"

GitHub Copilot: "Let me check your current workflow state..."

📋 Authentication TODOs (2/5 completed):
✅ Set up authentication routes
✅ Configure auth library integration  
⏳ Create login/register endpoints (next up)
⏳ Add JWT middleware for protected routes
⏳ Write authentication tests

You: "Use the existing auth library we have"

GitHub Copilot: "I see you already have `express-jwt` and `passport` in your dependencies. Let me update our approach to leverage these existing libraries and avoid redundant work..."

🔄 Updated todo: "Create login/register endpoints" → in_progress

*Configures passport strategies*
*Integrates with existing user model*  
*Creates streamlined auth endpoints using passport*

✅ Completed "Create login/register endpoints"

The existing auth library simplified our implementation significantly. Our persistent workflow state helped me avoid reprocessing the project setup...

🔄 Updated todo: "Add JWT middleware for protected routes" → in_progress

Quick Start

For Agent Mode (Recommended)

1. Install the extension from VS Code Marketplace
2. Chat with GitHub Copilot: Your AI now has todo_read and todo_write tools automatically
3. Start planning: Ask Copilot to create todos, track progress, or suggest next steps
4. Watch the magic: Todos appear in the Explorer tree view and sync with Copilot's memory

Try asking GitHub Copilot:

• "Create a todo list for implementing user authentication"
• "What should I work on next based on my current todos?"
• "Mark the database setup task as completed and suggest the next step"

Get Started

Quick Setup

1. Install from VS Code Marketplace
2. Open VS Code Chat and start planning with GitHub Copilot
3. Enable Auto-sync (optional): Settings > Extensions > Agent TODOs > Auto-inject

Try These Commands

Ask GitHub Copilot:

• "Create a todo list for implementing user authentication"
• "What should I work on next based on my current todos?"
• "Mark the database setup task as completed and suggest the next step"

Requirements

• VS Code 1.101.0 or higher
• GitHub Copilot extension

Contributing

Found a bug or have a feature request? Open an issue or submit a pull request.

Privacy Policy

Agent TODOs includes optional telemetry to help improve the extension:

• No personal data is collected - only anonymized usage patterns and error rates
• Todo content is never transmitted - your tasks remain private
• Disabled by default - requires explicit configuration to enable
• Follows VS Code telemetry standards - uses @vscode/extension-telemetry

Users can disable telemetry through VS Code's telemetry settings. See telemetry.json for details on data collected.

License

MIT License - see LICENSE for details.