Skip to main content

Mode Manager MCP: Server for managing VS Code .instructions.md files (GitHub Copilot persistent memory)

Project description

GitHub Copilot Memory Tool

Meet #remember -- Real Memory for You, Your Team, and Your AI

Install in VS Code Install in VS Code Insiders     License: MIT Python 3.8+

Mode Manager MCP is an AI-powered memory and context system for developers and teams. It lets you and your team “remember” important facts, preferences, and best practices—so your AI assistant always has the right context, and your team’s knowledge is never lost.

With Mode Manager MCP, you can:

  • Instantly store and retrieve personal, team, and language-specific knowledge.
  • Share onboarding notes, coding conventions, and project wisdom—right where you work.
  • Make your AI assistant smarter, more helpful, and always in sync with your workflow.

Why “Remember”? (Features & Benefits)

  • Personal AI Memory: Instantly store preferences, facts, and reminders for yourself—your AI assistant will always know your context.
  • Workspace (Team) Memory: Share best practices, onboarding notes, and team knowledge directly in the repo. New team members ramp up faster, and everyone stays on the same page.
  • Language-Specific Memory: Save and retrieve language-specific tips and conventions. Your assistant adapts to each language’s best practices automatically.
  • Natural Language Simplicity: Just say “remember…”—no config files, no YAML, no technical hurdles.
  • Smarter Coding, Fewer Repeated Questions: Your team’s memory grows over time, reducing repeated questions and ensuring consistent practices.- AI-Powered Memory Optimization: Automatically consolidate and organize your memories to keep them clean and efficient.  

 
Before this tool
"Hey Copilot, write me a Python function..."
Copilot: Gives generic Python code

After using remember
You: "Remember I'm a senior data architect at Oatly, prefer type hints, and use Black formatting"
Next conversation: "Write me a Python function..."
Copilot: Generates perfectly styled code with type hints, following your exact preferences
 

Ready to have Copilot that actually remembers you? Get started now!

Real-World Examples: Just Say It!

You don’t need special syntax—just talk to Copilot naturally. Mode Manager MCP is extremely relaxed about how you phrase things. If it sounds like something you want remembered, it will be!

 
Personal memory
You: I like detailed docstrings and use pytest for testing. (Copilot, keep that in mind.)


Team memory
You: We alw ays use the Oatly data pipeline template and follow our naming conventions. (Let’s make sure everyone remembers that.)


Language-specific memory You: For Python, use type hints and Black formatting. In C#, always use nullable reference types.
 

Get It Running (2 Minutes)

If you don't have Python installed, get it at python.org/downloads - you'll need Python 3.10 or higher.

Don't have uv yet? Install it with: pip install uv or see docs.astral.sh/uv

Option 1: One-Click Install (Easiest)

Click the badge for your VS Code version:

Install in VS Code Install in VS Code Insiders

That's it! The server will automatically start when you open VS Code.

Option 2: Manual Configuration

Add this to your .vscode/mcp.json or global mcp.json file:

{
  "servers": {
    "mode-manager": {
      "command": "uvx",
      "args": ["mode-manager-mcp"]
    }
  }
}

Or if you prefer pipx:

{
  "servers": {
    "mode-manager": {
      "command": "pipx",
      "args": ["run", "mode-manager-mcp"]
    }
  }
}

Option 3: Development / Testing (Latest from GitHub)

For the impatient who want the latest features immediately:

With uvx:

{
  "servers": {
    "mode-manager": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/NiclasOlofsson/mode-manager-mcp.git",
        "mode-manager-mcp"
      ]
    }
  }
}

With pipx:

{
  "servers": {
    "mode-manager": {
      "command": "pipx",
      "args": [
        "run",
        "--no-cache",
        "--spec",
        "git+https://github.com/NiclasOlofsson/mode-manager-mcp.git",
        "mode-manager-mcp"
      ]
    }
  }
}

This downloads and installs directly from GitHub every time - always bleeding edge!

Bonus: Interactive Onboarding

As a convenience, you can run the following prompt in VS Code to get started:

 
You: /mcp.mode-manager.onboarding
 

This will guide you through the onboarding process, set up your persistent memory, and ensure Copilot knows your preferences from the start.

For Development / Testing Latest from GitHub

To run the latest code directly from GitHub:

{
  "servers": {
    "mode-manager": {
      "command": "uv",
      "args": [
        "run",
        "--with",
        "git+https://github.com/NiclasOlofsson/mode-manager-mcp.git",
        "mode-manager-mcp"
      ],
      "env": {
        "_RESTART": "1"
      }
    }
  }
}

Under the Hood: How Memory Magic Happens

Mode Manager MCP is designed to make memory persistent, context-aware, and easy to manage—without you having to think about the details. Here’s how it works under the hood:

Memory Scopes

  • Personal Memory:
    Stored in a user-specific file (memory.instructions.md) in your VS Code prompts directory. This is your private memory—preferences, habits, and facts that follow you across all projects.

  • Workspace (Team) Memory:
    Stored in a workspace-level file (also memory.instructions.md, but in the workspace’s .github/instructions directory). This is shared with everyone working in the same repo, so team conventions and onboarding notes are always available.

  • Language-Specific Memory:
    Stored in files like memory-python.instructions.md, memory-csharp.instructions.md, etc. These are automatically loaded when you’re working in a particular language, so language tips and conventions are always at hand.

How Memory is Stored

All memory is saved as Markdown files with a YAML frontmatter header, making it both human- and machine-readable. Each entry is timestamped and neatly organized, so you can always see when and what was remembered. You never have to manage these files yourself—Mode Manager MCP automatically creates and updates them as you add new memories.

How Memory is Loaded

Here’s the magic: Mode Manager MCP writes and manages all your memory files, but it’s actually the new VS Code Copilot Chat that automatically loads them—every single turn. This deep integration means that, every time you send a message or ask Copilot for help, your user, workspace, and language memories are instantly available to the AI.

Language-specific memory is even smarter: it’s tied to file types using the applyTo property in the YAML frontmatter (for example, **/*.py for Python or **/*.cs for C#). This means you get the right tips, conventions, and reminders only when you’re working in the relevant language or file type—no clutter, just the context you need, exactly when you need it.

You never have to worry about context being lost between messages or sessions; your memory is always active and available. We’re simply leveraging this new, amazing VS Code feature to make your Copilot (and your team) smarter than ever.

No Special Syntax Needed

There’s no need to remember special commands or keywords—just talk naturally. Mode Manager MCP is flexible and understands a wide range of phrasing. You don’t have to say “workspace” to store team memory; it recognizes common alternatives like “project,” “repo,” or even just describing something as a team convention. Whether you’re making a personal note, a team guideline, or a language-specific tip, just say it in your own words—Mode Manager MCP figures out what you want to remember and where it belongs.

Contributing

Want to help make this better? The best contribution you can make is actually using it - your feedback and bug reports are what really drive improvements.

Of course, code contributions are welcome too! Check out CONTRIBUTING.md for development setup and guidelines. But seriously, just using it and telling us what works (or doesn't) is incredibly valuable.

License

MIT License - see LICENSE for details.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

File details

Details for the file iflow_mcp_niclasolofsson_mode_manager_mcp-2.0.2.tar.gz.

File metadata

  • Download URL: iflow_mcp_niclasolofsson_mode_manager_mcp-2.0.2.tar.gz
  • Upload date:
  • Size: 30.9 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.9.26 {"installer":{"name":"uv","version":"0.9.26","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Debian GNU/Linux","version":"13","id":"trixie","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}

File hashes

Hashes for iflow_mcp_niclasolofsson_mode_manager_mcp-2.0.2.tar.gz
Algorithm Hash digest
SHA256 8850f75e7c8a7fabba9b966e9adcb3bf6e6aefac9dddee4c28729499b7b446b5
MD5 dc20a7c7a32f3962a4d4e86271fa108b
BLAKE2b-256 95f4a5250e0f36e8643929b96e176abcd63cb5e57341df6138b8ed97c6d57ee4

See more details on using hashes here.

File details

Details for the file iflow_mcp_niclasolofsson_mode_manager_mcp-2.0.2-py3-none-any.whl.

File metadata

  • Download URL: iflow_mcp_niclasolofsson_mode_manager_mcp-2.0.2-py3-none-any.whl
  • Upload date:
  • Size: 37.2 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.9.26 {"installer":{"name":"uv","version":"0.9.26","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Debian GNU/Linux","version":"13","id":"trixie","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}

File hashes

Hashes for iflow_mcp_niclasolofsson_mode_manager_mcp-2.0.2-py3-none-any.whl
Algorithm Hash digest
SHA256 97d03432da2f45291bb06ae68971426088005a82b3a0e27c4eab7faa8051b132
MD5 83b45e1fde580b9061630ca01a0fa5d1
BLAKE2b-256 69230afef19dde02fb563578a9c541bf6ebf143aa56285967b14e893ad749485

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page