Skip to main content

Agents SDK CLI Reference

The Agents SDK provides a command-line interface (CLI) that lets you create, manage, and serve agents efficiently. This reference guide covers the various commands available in the CLI and how to use them.

Installation​

Before using the CLI, make sure you have installed the Agents SDK. If not, you can install it using pip:

pip install zetaalpha.rag-agents

CLI Commands​

Below is a list of all available commands in the Agents SDK CLI, along with examples and explanations.

init​

Initializes a new Agents SDK project. This command helps set up a new directory with the necessary files and structure for your agent project.

Usage:

rag_agents init [PROJECT_DIR]

Parameters:

  • PROJECT_DIR: (Required) The directory where the project will be initialized. If not provided, a wizard will prompt for the directory name.

Example:

rag_agents init my_agents_project

new​

Creates a new agent in the specified project directory. This command will generate the boilerplate code for a new agent, including the necessary configurations.

Usage:

rag_agents new [AGENT_NAME] [--project-dir PROJECT_DIR]

Parameters:

  • AGENT_NAME: (Required) The name of the agent to create. If not provided or if the name already exists in the project directory, a wizard will prompt for the agent name.
  • --project-dir: (Optional) The project directory where the agent will be created. If not provided, the current directory will be used if it's a valid project directory. Otherwise, a wizard will prompt for the project directory.

Example:

rag_agents new "my_new_agent" --project-dir my_agents_project

new-dependency​

Creates a new dependency in the specified project directory. This command will generate the necessary factory and class for the new dependency.

Usage:

rag_agents new-dependency [DEPENDENCY_NAME] [--project-dir PROJECT_DIR]

Parameters:

  • DEPENDENCY_NAME: (Required) The name of the dependency to create. If not provided or if the name already exists in the project directory, a wizard will prompt for the dependency name.
  • --project-dir: (Optional) The project directory where the dependency will be created. If not provided, the current directory will be used if it's a valid project directory. Otherwise, a wizard will prompt for the project directory.

Example:

rag_agents new_dependency "my_new_dependency" --project-dir my_agents_project

serve​

Starts the local REST API server for the agents project. Use this command to serve your agents as an API locally.

Usage:

rag_agents serve [PROJECT_DIR] [OPTIONS]

Parameters:

  • PROJECT_DIR: (Optional) The project directory where the dependency will be created. If not provided, the current directory will be used if it's a valid project directory. Otherwise, a wizard will prompt for the project directory.
  • --setup-src: (Optional) Path of the agent setup configuration. Only needed when using a non-standard project layout.
  • --secret-setup-src: (Optional) Path of the secret agent setup configuration. Only needed when using a non-standard project layout.
  • --reload: (Optional) Enable auto-reload. This will automatically reload the server when changes are detected in the code.
  • --host: (Optional) Host to listen on. Default is localhost.

Example:

rag_agents serve my_agents_project --host 0.0.0.0 --reload

dev​

Starts the Debugging Environment for the agents project. Use this command to run the development UI for testing and debugging agents.

Usage:

rag_agents dev [PROJECT_DIR] [OPTIONS]

Parameters:

  • PROJECT_DIR: (Optional) The project directory where the dependency will be created. If not provided, the current directory will be used if it's a valid project directory. Otherwise, a wizard will prompt for the project directory.
  • --setup-src: (Optional) Path of the agent setup configuration file. Only needed when using a non-standard project layout.
  • --secret-setup-src: (Optional) Path of the secret agent setup configuration file. Only needed when using a non-standard project layout.
  • --reload: (Optional) Enable auto-reload. This will automatically reload the UI when changes are detected in the code.
  • --storage-backend: (Optional) Storage backend to save the chat traces and evaluation results. Supported values are disk, s3, azure.
  • --storage-path: (Optional) Base path where the chat traces and evaluation results will be saved. This needs to be a valid path on the selected storage backend.

Example:

rag_agents dev my_agents_project --reload --storage-backend disk --storage-path ./agent-traces

version​

Prints the current version of the SDK.

Usage:

rag_agents version