Skip to content

Maticlib Logo

A high-performance Python Automation Library for creating intelligent AI agents.

License: MIT Downloads PyPI version

Maticlib is a developer-centric library built to orchestrate complex, stateful AI workflows. Whether you're building simple single-turn bots or massive multi-agent parallel graphs, Maticlib provides the primitives needed to ensure type safety, performance, and cross-provider consistency.

Key Features

Core Agent Framework

The MaticGraph workflow engine supports both stateful and stateless execution, conditional routing, and advanced loop prevention.

Unified LLM Integration

A single, consistent interface for OpenAI (Responses API), Google Gemini, and Mistral AI ensures portability and ease of use.

Advanced Telemetry

Track token usage across all modalities, including reasoning tokens for newer models, to optimize cost and performance.

Structured Response Parsing

Automatically parse and validate LLM outputs into Pydantic models, JSON, or XML. Maticlib handles the prompt engineering and extraction for you.

Type-Safe Orchestration

Deep integration with Pydantic ensures your workflow states and model responses are always valid and well-defined.

Quick Example

from maticlib.llm.openai import OpenAIClient

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

# Make a simple request
response = client.complete("Describe the benefit of graph-based AI orbits.")
print(client.get_text_response(response))

Maintained By

Maticlib is maintained and developed by Arvoh Software.