Volar - Plan, Review & Build with AI
Annoyed when AI starts touching unrelated code?
Getting nervous when you ask AI to write something maybe a little too complicated?
Pulling hair out when AI forgets your instructions?
Volar helps you guide your AI coding assistant more effectively:
- Defining Tasks: Write your idea and let AI fill in the details.
- Planning & Breakdown: Let your AI flesh out specifics, break down complex requests into manageable pieces, and plan implementation steps.
- Review & Refinement: You review the AI's plan, add context, and make adjustments before any code is written.
- Precise Execution: Instruct the AI to implement the approved plan, ensuring it sticks to the defined scope.
Volar is an orchestrator, preparing and sending detailed prompts to your chosen AI tool, and helping you manage the AI's output.
NOTE: Volar doesn't use AI directly. Volar creates prompts for your AI coding tool.
System Requirements
Volar is tested for macOS environments. While you can run Volar on other platforms, you may encounter unexpected behavior with some features.
🚀 Getting Started: Setup and Configuration
To use Volar, you first need to choose what AI tool to use it with. MCP(Model Context Protocol) is what Volar uses to share info with your AI tools (through calls like "update this task"). Choose the tool that you use in Volar's main task screen, and the MCP will be set up accordingly.

- Cursor: MCP will be automatically set up
- Windsurf: MCP will be automatically set up
- Claude Code: MCP will be automatically set up
- Other Tool: MCP needs to be set up manually
NOTE: You may need to manually refresh/enable the MCP config in Cursor and Windsurf: go to their MCP settings and verify VolarTaskServer
is enabled and running. You can click the ⚙️ icon next to tool dropdown menu to go to settings. Then search "MCP" to find the right setting.
For other tools, as long as they support MCP, Volar can be set up to work with them. You can find the Volar MCP url next to tool dropdown menu.

MCP Setup for Cline
- Copy the MCP url above Volar's task list.
- Go to Cline's MCP setup tab.
- Use the following values:
- Server Name:
VolarTaskServer
- Server URL:
http://localhost:<port shown>/sse
(the copied url)
💡 Core Features & Usage
Once Volar is configured, you can start planning, reviewing, and building with your AI!
Task View & Task Management
The Volar extension adds a new view to your VS Code sidebar where you can manage your tasks.


- View Task: Click on any task to see its details.
- Create Tasks: Write down a task title and hit "Enter" to see the new task.
- Edit Task Details: In task detail view, click the "✏️" (pencil) emoji to edit the task content.
- AI-Assisted Task Updates (with MCP): If you've set up MCP, you can also ask your AI agent (e.g., Cursor) to update tasks directly for you through chat.
Taking Actions on Tasks with AI
For each task, you can ask your AI to perform specific actions.

If you are using Claude Code, the task action prompt will be sent to terminal automatically. Hit enter
to execute the command.
If you are using Cursor/Windsurf/etc, the prompt will be copied to your clipboard, and you can paste it into your coding tool's chat window.

There are three key actions:
Plan:
- What it does: Provide a concise task title or a brief description. Volar will then prompt your AI to analyze your codebase (if relevant) and generate a detailed implementation plan.
- When to use it: For any new idea or feature. This is your first step to get the AI to articulate its understanding and proposed solution.
- Outcome: The AI's plan will populate the task details. Review this plan carefully. Make changes as needed.
Breakdown:
- What it does: If a task or its plan seems too complex, ambiguous, or large for the AI to handle in one go, use "Breakdown". Volar will prompt your AI to break the current task into smaller, more manageable sub-tasks.
- When to use it: When a task feels overwhelming, or the initial "Plan" is too high-level.
- Outcome: New sub-tasks will be created. You can recursively use "Plan" and "Breakdown" on these sub-tasks until each one is small and clear enough for reliable execution.
Execute:
- What it does: After you have reviewed and are satisfied with a task's plan (either an initial plan or one refined after breakdowns), use "Execute". Volar will prompt your AI to write the code according to that specific plan.
- When to use it: When the plan is clear, detailed, and you're confident the AI can implement it correctly. For very simple, self-contained changes, you might execute directly.
- Outcome: The AI will attempt to make the code changes according to the task content.
⚠️ Known Issues & Troubleshooting
- Your AI coding tool cannot get task info from Volar: Please check the MCP settings of your AI coding tools (Cursor/Windsurf/etc) and verify the status of
VolarTaskServer
is enabled and running, and the server URL matches the MCP URL
shown in the Volar extension, next to the Tool dropdown menu.