maticlib 0.1.7

A Python Automation Library for Creating Agents

A high-performance Python Automation Library for Creating Agents.

Maticlib is a developer-centric library designed to build complex, stateful AI workflows with ease. Whether you need a simple chat completion or a multi-node parallel execution graph, Maticlib provides the primitives to build it with a focus on type safety and performance.

---

Key Features

• Core Agent Framework: Pure-Python graph workflow engine (MaticGraph) supporting both stateful and stateless execution.
• Unified LLM Integration: Single interface for OpenAI (modern Responses API), Google Gemini, and Mistral AI.
• Complex Orchestration: Built-in support for conditional routing, parallel node execution, and loop prevention.
• Data Integrity: Deep Pydantic integration for standardized response models and type-safe workflow states.
• Advanced Telemetry: Detailed token usage tracking, including modalities and reasoning tokens for newer models.

---

Documentation

For complete documentation, visit: https://arvohsoft.github.io/maticlib/

• Getting Started
• API Reference
• Example Gallery

---

Installation

pip install maticlib

---

Quick Start

from maticlib.llm.openai import OpenAIClient

# Initialize client (uses OPENAI_API_KEY from environment)
client = OpenAIClient()

# Make a request
response = client.complete("Explain quantum computing in one sentence.")
print(client.get_text_response(response))

---

Roadmap

Core Library

• Google Gemini integration
• Mistral AI integration
• OpenAI Responses API integration
• Graph-based workflow engine
• Parallel node execution
• Automated documentation system
• Anthropic Claude integration
• Streaming support for all providers

Advanced Features

• Unified tool/function calling interface
• Multi-agent collaboration protocols
• MCP (Model Context Protocol) support
• Workflow persistence & checkpointing

---

Project Leadership

Maticlib is developed and maintained by Arvoh Software.

• Lead Maintainer: Anubroto Ghose
• Organization: Arvoh Software
• Contact: [e-mail]

---

License

This project is licensed under the MIT License - see the LICENSE file for details.