Agent Swarm Protocol Clients

This directory contains client applications for interacting with the Agent Swarm Protocol orchestrator.

Terminal Client

The Terminal Client provides a command-line interface for interacting with the orchestrator:

• List available agents
• Send tasks to agents
• Check task status
• Execute workflows

Installation

# From the clients directory
npm install
npm link  # Installs the client globally as 'asp-cli'

# Or from the parent directory
npm run client:install

Usage

You can run the client in two ways:

# Using the global command (if linked)
asp-cli

# Or directly with Node
node terminal-client.js

# Or from the parent directory using npm
npm run client:terminal

Configuration

The client connects to the Orchestrator's client WebSocket port. By default, it uses:

• ws://localhost:3001

You can configure the connection URL by setting the ORCHESTRATOR_CLIENT_URL environment variable in the .env file in the parent directory.

Available Commands

Once the client is running, the following commands are available:

• agents - List all available agents
• task - Send a task to an agent
• status - Check the status of a task
• workflow - Execute a workflow
• help - Show the help menu
• exit or quit - Exit the client

Example: Sending a Task to the Conversation Agent

> agents
# View the available agents

> task
# Select the conversation-agent
Enter agent name: conversation-agent

# Enter the task data
Task data: {"message": "Hello, how are you?", "conversationId": "test-123"}

# The agent will process the task and send back a response

Example: Executing a Workflow

> workflow
Enter workflow name: example-workflow

# Enter any additional options
Options: {"initialMessage": "Start the process", "parameters": {"key": "value"}}

# The workflow will execute and return its results

Creating New Clients

You can use the Terminal Client as a template for creating new clients. The key components are:

1. Establish a WebSocket connection to the orchestrator's client port
2. Send messages with the correct format for listing agents, creating tasks, etc.
3. Process the responses from the orchestrator

See the Agent Swarm Protocol documentation for more details on the message formats and API.