Skip to main content

​
Overview

This guide covers the essentials: create an agent, test it in the playground, add knowledge retrieval, and configure access. Expected time: 10 minutes.
Prerequisites: Atthene Agents Studio account, basic YAML familiarity

​
Create Your First Agent

​
Access the Playground

Navigate to Playground in the sidebar. You’ll see:
  • Left panel: YAML configuration editor
  • Right panel: Agent Preview for testing

​
Load a Template

Click “Load Config” and select “Basic Dynamic Agent Test” to start with a working configuration. 
name: "Basic Dynamic Agent Test"
description: "Test configuration for basic dynamic agent loading"
save_messages: true
persistent_state: true

agents:
  - name: "test_basic_agent"
    agent_type: "llm_agent"
    description: "A simple sessional agent for testing dynamic loading"
    system_prompt: "You are a helpful AI assistant."
    llm_config:
      temperature: 1.0
      max_tokens: 500
    memory_config: {}
    knowledge_bases: []
    streaming_config:
      show_output_to_user: true

edges:
  - from: "__start__"
    to: "test_basic_agent"
  - from: "test_basic_agent"
    to: "__end__"

entry_point: "test_basic_agent"

​
Test the Agent

Click on Load Configuration. Verify the Agent Preview shows “Connected”, then send a test message. The agent uses the loaded configuration to respond.

​
Add Knowledge Retrieval

Knowledge retrieval requires creating a chain of entities: Data Source → Collection → Knowledge Base → Retrieval Method.

​
Create Data Source

Navigate to Knowledge Base → Data Sources tab. Click ”+ Add Data Source” and upload your files. Wait for processing to complete.

​
Create Collection

Go to Collections tab. Click ”+ Create Collection” and add your data source(s) to it. Collections group related documents.

​
Create Knowledge Base

Navigate to Knowledge Bases tab. Click ”+ Create Knowledge Base” and select your collection. This creates a searchable index.

​
Configure Retrieval Method

In Retrieval Methods tab:
  1. Click ”+ Create Retrieval Method”
  2. Select your knowledge base
  3. Configure search parameters (similarity threshold, max results, etc.)
  4. Copy the generated configuration string

​
Connect to Agent

Paste the copied configuration into your YAML: 
agents:
  - name: "test_basic_agent"
    knowledge_bases: ["kb_uuid_from_copied_config"]  # Paste here
Load the new configuration and test with questions about your uploaded content. The agent retrieves and references relevant information based on your retrieval method settings.

​
Save and Deploy

​
Persist Your Agent

Click “Save as Agent” in the editor, provide a name.
Saved agents appear in Agent Overview and can be accessed via API or embedded.

​
Deploy a Version

Saving creates a new version. To make it active for API/connections: Agent Overview → Select agent → Settings → Versions → Click Deploy on desired version
The version with the “current” banner is actively deployed and handles all API requests and connections.

​
Access Your Agent from Software or Websites

Ready to integrate your agent into your applications? You can access your deployed agents through REST APIs, WebSocket connections, or embed them directly into websites and SharePoint. This enables seamless integration with existing software systems and user interfaces.

​
Next Steps

Connecting Your Agent

Learn how to access your agent from software or websites using REST API, WebSocket, and embedding

YAML Configuration

Dive deeper into the agent framework with full configuration options and syntax

Agent Types

Learn about different agent types and architectures

Knowledge Bases

Connect your agents to domain-specific information

Feedback

Questions or suggestions? Click “Give Feedback” in the platform or email [email protected]