Setup AgentComet Locally

Run your own private agent registry and orchestration studio on your local machine using Docker and our SDK.

Prerequisites

Python 3.8 or higher
Docker installed and running
An LLM provider (Ollama, OpenAI, Gemini, or Anthropic)

1. Install the SDK

Install the core AgentComet framework via pip to start building agents.

bash

pip install agentcomet

2. Start AgentComet Studio

Run the local registry using Docker to enable version control and agent sharing.

Linux / macOS

bash

docker pull vaibhavhaswani/agentcomet-studio:latest
docker run -p 3451:3451 -v $(pwd)/data:/app/data vaibhavhaswani/agentcomet-studio:latest

Windows (PowerShell)

powershell

docker run -p 3451:3451 -v ${PWD}/data:/app/data vaibhavhaswani/agentcomet-studio:latest

3. Configure Connection

Point your SDK to your local Studio instance using environment variables or in-code settings.

Environment Variables

bash

export AGENTCOMET_URL=http://localhost:3451
export AGENTCOMET_KEY=your-local-key

Python Initialization

python

from agentcomet import Settings, Agent

Settings.init(
    AGENTCOMET_URL="http://localhost:3451",
    AGENTCOMET_KEY="your-local-key"
)

try:
    push_res = agent.push(repo="vhx/assistant", version="auto", create=True)
    print("Push response:", push_res)

    downloaded_agent = Agent.pull(repo="vhx/assistant", version="latest")
    print("Downloaded agent:", downloaded_agent.name)
except Exception as e:
    print("Local Server Interaction Error:", e)

Need more help? Check out the GitHub repository for full documentation.