Getting Started

Welcome to Agnet! This guide will walk you through everything you need to know to get started.

Docker Setup

For the full experience, including the web UI and isolated environments, we recommend using Docker. You can download it from the official Docker website.

Python Virtual Environment

Before installing dependencies, it's a good practice to create a virtual environment. Create one by running:

bash

python3 -m venv venv

Then, activate it:

bash
# On Windows (Git Bash or CMD)
.\venv\Scripts\activate

# On macOS/Linux
source venv/bin/activate

Installation

Create a virtual environment and install Laddr (CLI, core, API):

bash

pip install laddr

Or develop against this repository (editable):

bash

pip install -e lib/laddr

Create a new project

bash

laddr init my_agent_system
cd my_agent_system

The project includes agents/, workers/, a Dockerfile, docker-compose.yml, and a main.py runner.

Set API keys

To quickly get started, add the following env vars to a .envfile in your project root:

bash
# .env
GEMINI_API_KEY=your_gemini_api_key
SERPER_API_KEY=your_serper_api_key

Note: one of our tool templates (web search) requires a Serper API key; Gemini is used for LLM integrations. Set both before running the stack.

Run the Stack (Docker)

bash

laddr run dev -d

bash

docker compose up -d

The framework uses Docker under the hood, so both commands start the same local stack. Open the dashboard at http://localhost:5173 and the API at http://localhost:8000.

Add an agent and a tool

bash
laddr add agent researcher --role "Researcher" --goal "Find facts" --llm-model gemini-2.5-flash
laddr add tool web_search --agent researcher --description "Search the web"

Quick Run

bash

laddr run coordinator '{"topic": "Latest AI agent trends"}'

Note: this quick run starts a single coordinator worker. To run your entire stack locally without Docker, see the Local Setup section.