Skip to main content

Agents

An Agent in OGX is a powerful abstraction that allows you to build complex AI applications.

The OGX agent framework is built on a modular architecture that allows for flexible and powerful AI applications. This document explains the key components and how they work together.

Core Concepts​

1. Agent Configuration​

Agents are configured using the AgentConfig class, which includes:

  • Model: The underlying LLM to power the agent
  • Instructions: System prompt that defines the agent's behavior
  • Tools: Capabilities the agent can use to interact with external systems
  • Guardrails: Optional moderation checks via the Responses provider
from ogx_client import Agent

# Create the agent
agent = Agent(
    ogx_client,
    model="meta-llama/Llama-3-70b-chat",
    instructions="You are a helpful assistant that can use tools to answer questions.",
    tools=["builtin::code_interpreter", "builtin::file_search"],
)

2. Sessions​

Agents maintain state through sessions, which represent a conversation thread:

# Create a session
session_id = agent.create_session(session_name="My conversation")

3. Turns​

Each interaction with an agent is called a "turn" and consists of:

  • Input Messages: What the user sends to the agent
  • Steps: The agent's internal processing (inference, tool execution, etc.)
  • Output Message: The agent's response
from ogx_client import AgentEventLogger

# Create a turn with streaming response
turn_response = agent.create_turn(
    session_id=session_id,
    messages=[{"role": "user", "content": "Tell me about Llama models"}],
)
for log in AgentEventLogger().log(turn_response):
    log.print()

4. Steps​

Each turn consists of multiple steps that represent the agent's thought process:

  • Inference Steps: The agent generating text responses
  • Tool Execution Steps: The agent using tools to gather information
  • Guardrails Steps: Moderation checks being performed

Agent Execution Loop​

Refer to the Agent Execution Loop for more details on what happens within an agent turn.