Atomic Agents

Atomic Agents is a Python framework for building modular, structured AI agents on top of Instructor and Pydantic. It talks to any OpenAI-compatible endpoint, so you can power every agent with Qubax by passing it a client pointed at the Qubax base URL.

Install

pip install atomic-agents

from openai import OpenAI
import instructor
from atomic_agents import AtomicAgent, AgentConfig, BasicChatInputSchema

# Point Instructor at Qubax — everything else is stock Atomic Agents
client = instructor.from_openai(
    OpenAI(
        base_url="https://api.qubax.ai/v1",
        api_key="YOUR_API_KEY",
    )
)

agent = AtomicAgent[BasicChatInputSchema, BasicChatInputSchema](
    config=AgentConfig(
        client=client,
        model="gpt-5",
    )
)

response = agent.run(
    BasicChatInputSchema(chat_message="Explain quantum entanglement in one sentence.")
)
print(response.chat_message)

Setup steps

1. 1

   Install the framework

   Run pip install atomic-agents. The openai SDK (with OpenAI support) is included by default; no extra provider extras are needed for Qubax.

2. 2

   Create the client

   Build an instructor client from an OpenAI client configured with base_url="https://api.qubax.ai/v1" and api_key set to your Qubax key (starts with qbx_live_).

3. 3

   Pass it to your agent

   Give the client to AgentConfig along with a model such as gpt-5. Every agent run, chain, and pipeline now routes through Qubax.