Skip to main content
Glama

@arizeai/phoenix-mcp

Official
by Arize-ai
README.md3 kB
# Computer Use Agent Examples ## Overview This repository contains examples for building a **Computer Use (Operator) Agent** using **LangGraph** and **Anthropic**. The agent leverages Anthropic's **beta APIs** to perform various computer-related tasks, including executing Bash commands, editing files, and performing system operations. **⚠️ Important:** Since the agent can execute system-level operations, it is recommended to run it inside a **virtual machine (VM)** for safety. To facilitate this, we provide **Docker files** to set up an **Ubuntu virtual machine** and **noVNC** for accessing the VM via a web browser. **💡 Note:** All the tools used in this example are **Ubuntu compatible**. --- ## Features - **UI:** Built with Gradio app - **Computer Use Agent Workflow:** Built using **LangGraph** - **Anthropic Model Integration:** Uses **Anthropic's Computer Use beta models** - **Prebuilt Tools:** Includes tools for **Bash execution, file editing, and system operations** - **Instrumentation with Phoenix:** Auto-instrumentation with OpenInference decorators to fully instrument the agent - **End-to-end tracing:** Track agent performance using Phoenix --- ## Requirements - **Anthropic API Key** - **Python 3.x** - **LangGraph library** - **Phoenix for instrumentation** - **Gradio for UI** --- ## Installation 💡 **No need to install dependencies manually!** The required libraries are already included in the **Docker image**. Just Run the docker image with docker command, open a web browser and access the Ubuntu environment using **noVNC**. --- ## Usage ### 1. Start the Virtual Machine Using Docker First, build the Docker image: ```sh docker build -t computer-use-agent:1.0 . ``` Run the container: ```sh docker run -v $HOME/.anthropic:/home/computeruse/.anthropic \ -p 5900:5900 -p 7860:7860 -p 6080:6080 -p 8080:8080 \ -it computer-use-agent:1.0 ``` ### 2. Access the Virtual Machine Once the container is running, access the Ubuntu environment using **noVNC**: - Open a web browser and visit: ``` http://localhost:8080 ``` ### 3. Input all the required configurations in the webpage * **Anthropic API Key** * **Phoenix API Key** * **Phoenix Collector Endpoint** * **Phoenix Project Name** ### 4. Run the Agent - Provide input commands, and the agent will execute them inside the **Ubuntu virtual machine**. - Monitor system actions via the **VNC interface**. --- ## File Structure - **`app.py`** – Main script for running the agent. - **`agent.py`** – Agent script. - **`tools/`** – Contains utilities for computer, bash, edit file tools. - **`requirements.txt`** – List of required dependencies (**already included in the Docker image**). --- ## Notes - Ensure you run the agent inside a **virtual machine** to prevent unintended system modifications. - **All tools used in this model are Ubuntu compatible.**

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/Arize-ai/phoenix'

If you have feedback or need assistance with the MCP directory API, please join our Discord server