Mirror VS

Your AI pair programmer.

Code, architect, debug, and ask questions naturally. Mirror VS adapts to how you work.

What is Mirror VS?

Mirror VS is your AI-powered development sidekick. Think of it as a brilliant collaborator who never sleeps — ready to write code, plan architectures, squish bugs, answer questions, or automate the boring stuff. All without leaving your editor.

Whether you're spinning up a new project, refactoring a gnarly codebase, or trying to understand why that test won't pass, Mirror VS has your back.

📖 Full documentation at reflectai.in

What can it do?

• Write code from plain English descriptions and specs
• Architect systems — plan migrations, design APIs, document decisions
• Refactor and debug existing code with surgical precision
• Answer questions about your codebase instantly
• Automate repetitive tasks so you can focus on what matters
• Connect to MCP servers for extended superpowers
• Index your codebase for context-aware file suggestions and semantic search
• Manage tasks with structured todo lists, checkpoints, and subtask delegation

Modes for every moment

Mirror VS adjusts its behavior to match what you need:

Each mode remembers its own model selection, so you can assign different models per mode (e.g., Gemini for architecting, Claude for coding) — and Mirror VS switches automatically when you change modes.

🔗 Full guide to using modes

Getting started

1. Install

Install from the VS Code Marketplace or Open VSX.

2. Connect an AI provider

Mirror VS supports multiple providers. On first launch, the onboarding wizard will guide you through choosing and configuring one:

You can also configure multiple provider profiles and switch between them — useful for separating work and personal API keys or assigning different models to different modes.

🔗 Full provider setup guide

3. Start a task

Open the command palette (Cmd+Shift+P) and run Mirror VS: Start New Task, or click the chat icon in the activity bar. Tell Mirror what you need — in your own words. That's it.

4. Try slash commands

Speed up common workflows with slash commands:

• /code — Switch to Code mode
• /architect — Switch to Architect mode
• /ask — Switch to Ask mode
• /debug — Switch to Debug mode
• /todo — Show or update the task todo list
• /clear — Clear the current task

Custom modes

One of Mirror VS's most powerful features is the ability to create custom modes — specialized personas with their own instructions, tool access, and model preferences.

How to create a custom mode

1. Create a .mirrormodes file in your project root (team-wide) or use Mirror VS > Custom Modes in settings (personal)
2. Define your mode with a slug, name, role definition, and tool configuration:

{
	"customModes": [
		{
			"slug": "reviewer",
			"name": "Code Reviewer",
			"roleDefinition": "You are a meticulous code reviewer. Focus on security, performance, and readability.",
			"groups": ["read", ["edit", { "fileRegex": "*.ts", "description": "TypeScript files only" }]]
		}
	]
}

Your custom mode will appear in the mode selector dropdown alongside built-in modes.

🔗 Full custom modes guide

MCP Server integration

Mirror VS supports the Model Context Protocol (MCP) — an open standard for connecting AI assistants with external tools and data sources.

Popular MCP servers to try

How to add an MCP server

1. Open Mirror VS settings (Cmd+,)
2. Navigate to the MCP Servers section
3. Click Add Server and choose the transport type:
   • Stdio: For local servers (Node.js, Python, Go binaries)
   • SSE: For remote HTTP servers
4. Configure the command, arguments, and environment variables

Example — adding a filesystem server:

{
	"mcpServers": {
		"filesystem": {
			"command": "npx",
			"args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/allowed/dir"],
			"disabled": false,
			"autoApprove": []
		}
	}
}

🔗 MCP server setup docs | Official MCP specification

Code indexing & semantic search

Mirror VS can index your codebase for fast, context-aware file retrieval. When enabled, it:

• Builds embeddings for your project files
• Finds relevant files based on natural language queries
• Surfaces context automatically as you work

Configure it in Settings > Code Indexing. Supports OpenAI, Gemini, Ollama, Mistral, AWS Bedrock, OpenRouter, Vercel, and more.

Key features at a glance

License

Apache 2.0 © 2026 ReflectAI