Microsoft Amplifier - Specialized AI Agents

Amplifier: From Installation to Utilization, AI Development Environment Enhancement Guide

#

AI coding assistants have now become an essential tool for many developers. However, when sessions get long, they can lose context, or hit limits where it’s difficult to apply consistent guidelines across multiple projects. Amplifier is an enhanced development environment designed to solve exactly these problems.

In this guide, we will learn step-by-step what Amplifier is, how to install it, and how to safely apply it to existing projects to boost development productivity.

What is Amplifier?

#

Amplifier is an integrated development environment that supercharges general AI coding assistants with expert teams, accumulatable knowledge, and powerful automation systems. Instead of starting from scratch in every new session, Amplifier instantly provides proven patterns, specialized agents, and workflows to help implement complex solutions with minimal instruction.

• Over 20 Specialized Agents: Provides agents specialized for specific tasks such as bug-hunter (debugging), zen-architect (design), etc.
• Knowledge Base System: Acquires project documents, notes, etc., and accumulates them as permanent knowledge that the AI can query.
• Parallel Worktree System: Build and test multiple ideas or solutions simultaneously.
• Automation Tools: Code quality checks, automatic saving and restoration of conversation history, etc.

Step 1: Installation and Environment Setup (QuickStart)

#

Check and install the necessary tools to use Amplifier.

Check Installation Status

#

• python3 --version # 3.11 or higher required
• uv --version # Any version
• node --version # Any version
• pnpm --version # Any version
• git --version # Any version

Install Necessary Tools

#

Mac

brew install python3 node git pnpm uv

Ubuntu/Debian/WSL

# System packages
sudo apt update && sudo apt install -y python3 python3-pip nodejs npm git

# pnpm
npm install -g pnpm
pnpm setup && source ~/.bashrc

# uv (Python Package Manager)
curl -LsSf https://astral.sh/uv/install.sh | sh

Windows

1. Install WSL2.
2. Run the Ubuntu commands above in the WSL environment.

Amplifier Setup

#

# 1. Clone repository and move to directory
git clone https://github.com/microsoft/amplifier.git
cd amplifier

# 2. Install Python dependencies
make install

# 3. Activate virtual environment
source .venv/bin/activate # Linux/Mac/WSL
# .venv\Scripts\Activate.ps1 # Windows PowerShell

Step 2: Applying to Projects (Using Symbolic Links)

#

The biggest advantage of Amplifier is that you can use all features while keeping the original project files intact. This is made possible through ‘Symbolic Links’ (shortcuts).

1. Move to the Amplifier directory. (Skip this step if you are already in the amplifier directory)

   cd ~/amplifier

2. Create a symbolic link for the project you want to work on inside the ai_working/ directory.

   # Example: Linking my project located at /path/to/my/project
   ln -s /path/to/my/project ai_working/my-project-link

3. Run Claude Code in the Amplifier directory.

   claude

4. Specify the project to work on in the Claude Code session.

I’m working in ai_working/my-project-link, and using the capabilities from amplifier.

Important: Modifying, creating, or deleting files via the symbolic link directly reflects on the original project. Therefore, it is always safe to track changes with Git.

Step 3: Utilizing Key Features

#

Calling Specialized Agents

#

Instead of simple requests, you can specify agents with specific roles to get more professional results.

“Use the zen-architect agent to design my application’s caching layer.”
“Deploy bug-hunter to find why my login system is failing.”

Building and Utilizing Knowledge Base

#

When you add project-related documents and run the make knowledge-update command, Amplifier analyzes the content to extract concepts and relationships.

As this process repeats, the knowledge available to the AI compounds, leading to a deeper understanding of the project. This is why it gets “smarter over time”.

1. Update Knowledge Base: (Run after adding documents like docs/, specs/ to the project)

   make knowledge-update

2. Query Accumulated Knowledge:

   make knowledge-query Q="authentication patterns"

FAQ

#

Q1: What is the difference between symbolic links and file copying?

#

• Symbolic Link: A ‘shortcut’ pointing to the original file. It takes up almost no disk space, and modifications through the link immediately change the original.
• Copy: Completely duplicates the file. Uses double the disk space, and the original and copy do not affect each other.

Q2: Can I connect multiple projects simultaneously?

#

Yes, it is possible. You can create multiple symbolic links in the ai_working/ directory.

ln -s ~/project-a ai_working/project-a
ln -s ~/project-b ai_working/project-b

Q3: How do I work on a specific project when multiple projects are connected?

#

Always run claude from the ~/amplifier directory, and clearly specify the project to work on at the start of the session.

“I’m working only in ai_working/project-a/. Please don’t modify other projects.”

Caution: If you move directly to the directory with the symbolic link (~/amplifier/ai_working/project-a) and run claude, you cannot use Amplifier’s features (.claude/ settings).

Q4: How do I safely remove a symbolic link?

#

You just need to remove the link file with the rm command. The original project remains safe.

# Remove link only (Safe)
rm ai_working/my-project-link

# ⚠️ NEVER use this (Contents of the original directory may be deleted)
# rm -rf ai_working/my-project-link/

Additional Resources

#

• Amplifier README
• Agent List - All available agents
• Claude Code Official Documentation
• Symbolic Link Explanation