Karo Framework

Karo is a modular Python framework designed for building robust and maintainable AI agent applications. It emphasizes:

• Modularity: Construct agents from reusable components (core logic, memory, tools).
• Predictability: Define clear agent interactions using Pydantic schemas.
• Memory: Equip agents with persistent memory using vector stores like ChromaDB.
• Extensibility: Easily integrate various LLM providers and custom tools.
• Control: Maintain fine-grained control over agent behavior and data flow.

Inspired by frameworks like Atomic Agents, Karo aims to provide developers with the building blocks needed for creating sophisticated, reliable AI agents.

Key Features

• Schema-Driven Development: Use Pydantic for defining agent inputs, outputs, and tool parameters.
• Provider Abstraction: Pluggable interface for different LLM providers (OpenAI included).
• Persistent Memory: Built-in MemoryManager using ChromaDBService for storing and retrieving relevant context.
• Tool Integration: Define and integrate custom tools for agents to use.
• Pythonic Control Flow: All agent logic and orchestration is written in Python.

Installation

pip install karo

See the full Installation Guide for details on setting up dependencies and environment variables (like OPENAI_API_KEY).

Quickstart

Here's a minimal example of creating a basic agent:

import os
from dotenv import load_dotenv
from karo.core.base_agent import BaseAgent, BaseAgentConfig
from karo.providers.openai_provider import OpenAIProvider, OpenAIProviderConfig
from karo.schemas.base_schemas import BaseInputSchema, BaseOutputSchema

# Load API keys from .env file in your project root
load_dotenv()

# Configure provider (requires OPENAI_API_KEY in .env)
provider_config = OpenAIProviderConfig(model="gpt-4o-mini")
provider = OpenAIProvider(config=provider_config)

# Configure agent
agent_config = BaseAgentConfig(provider=provider)
agent = BaseAgent(config=agent_config)

# Run agent
input_data = BaseInputSchema(chat_message="Explain the concept of AI agents briefly.")
result = agent.run(input_data)

if isinstance(result, BaseOutputSchema):
    print(f"Agent: {result.response_message}")
else:
    print(f"Error: {result.error_message}")

For more detailed examples, including memory and tool usage, check out the Quickstart Guide.

Documentation

Full documentation, including guides and API references, can be found in the docs/ directory (or at the hosted documentation site - link TBD).

• Installation
• Quickstart Guide
• (More guides coming soon)

License

Copyright 2025 Mensterra

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.