Computer Use Agent

An autonomous desktop automation agent powered by Claude's Computer Use API and LangGraph. The agent takes screenshots, analyzes them with Claude, and executes mouse and keyboard actions to complete tasks on your computer.

Example: Creating a PowerPoint presentation about dogs.

How to Run

Prerequisites

• Python 3.13+
• uv package manager
• Anthropic API key with Computer Use access
• Langfuse

Installation

git clone https://github.com/chr-peters/computer-use-agent.git
cd computer-use-agent
uv sync

Configuration

Create a .env file with your credentials:

ANTHROPIC_API_KEY=your-api-key
ANTHROPIC_BASE_URL=https://api.anthropic.com  # optional, for custom endpoints

LANGFUSE_PUBLIC_KEY=your-public-key
LANGFUSE_SECRET_KEY=your-secret-key
LANGFUSE_HOST=https://cloud.langfuse.com  # or your self-hosted instance

Running

Edit the task in main.py to specify what you want the agent to do, then run:

uv run run-workflow

To visualize the agent's graph structure:

uv run visualize