sofia-agent 0.1.7

SOFIA: Simple Orchestrated Flow Intelligence Agent

Simple Orchestrated Flow Intelligence Agent

SOFIA is an open-source, configurable multi-step agent framework for building advanced LLM-powered assistants. Define your agent's persona, tools, and step-by-step flows in Python or YAML—perfect for conversational, workflow, and automation use cases.

Features

• Step-based agent flows: Define agent behavior as a sequence of steps, each with its own tools and transitions.
• Persona-driven: Easily set the agent's persona for consistent, branded responses.
• Tool integration: Register Python functions as tools for the agent to call.
• YAML or Python config: Configure agents via code or declarative YAML.
• OpenAI and custom LLM support
• Session management: Save and resume conversations.
• Extensible: Build your own tools, steps, and integrations.
• Interactive CLI: Bootstrap new agents with sofia init (install with [cli] extra).

Installation

From PyPI

pip install sofia-agent

With CLI support

pip install sofia-agent[cli]

From Source

git clone https://github.com/chandralegend/sofia.git
cd sofia
poetry install

Usage

CLI: Bootstrap a New Agent

sofia init

This will interactively guide you to create a config YAML and starter Python file for your agent.

Python API Example

from sofia_agent import *
from sofia_agent.llms import OpenAIChatLLM

def get_time():
    from datetime import datetime
    return f"Current time: {datetime.now()}"

steps = [
    Step(
        step_id="start",
        description="Greet and offer to tell the time.",
        available_tools=["get_time"],
        routes=[Route(target="end", condition="User is done")],
    ),
    Step(
        step_id="end",
        description="Say goodbye.",
    ),
]

llm = OpenAIChatLLM()
agent = Sofia(
    name="clockbot",
    llm=llm,
    steps=steps,
    start_step_id="start",
    tools=[get_time],
    persona="You are a friendly clock assistant.",
)
sess = agent.create_session()
# ... interact with sess.next(user_input)

YAML Config Example

See examples/config.barista.yaml for a full-featured barista agent.

Configuration

• Persona: Set in YAML or Python for consistent agent style.
• Steps: Each step defines available tools, description, and routes to other steps.
• Tools: Python functions registered with the agent.

Example: Barista Agent

A full example is provided in examples/barista/barista.py and examples/config.barista.yaml.

To run the Barista agent:

cd examples/barista
export OPENAI_API_KEY=your-api-key-here
python barista.py
# or
python barista_with_config.py

Example: Financial Planning Assistant

A production-ready example of a Financial Planning Assistant is available in examples/prod-ready/. This example demonstrates:

• Budget planning and expense tracking
• Savings goal management
• Financial health assessment
• Docker containerization with optional Redis session management
• Production-ready configuration

To run the Financial Planning Assistant:

OPENAI_API_KEY=your-api-key-here docker compose up

Contributing

Contributions are welcome! Please open issues or pull requests on GitHub.

License

MIT License. See LICENSE.

Acknowledgements

• Inspired by the open-source LLM community.
• Built with ❤️ by contributors.