<p align="center"> <img src=".github/assets/peargent.png" alt="Peargent Logo"> </p>

Peargent

A modern Python framework for building intelligent AI agents with simplicity at its core.

Features

• Advanced Tracing - Track every action, decision, and API call with detailed telemetry and database persistence
• Smart History Management - Built-in conversation history with intelligent context windowing and buffer management
• Multi-LLM Support - Seamlessly switch between OpenAI, Anthropic, Groq, Gemini, and more
• Type-Safe Tools - Pydantic-powered tool system with automatic validation
• Agent Pools - Run multiple agents concurrently with shared context
• Streaming Support - Real-time streaming responses with event handling
• Cost Tracking - Monitor token usage and costs across all LLM providers

Installation

pip install peargent

Quick Start

from peargent import create_agent
from peargent.models import groq, anthropic, openai

# Use any model provider
agent = create_agent(
    name="assistant",
    persona="You are a helpful AI assistant.",
    model=anthropic("claude-3-5-sonnet-20241022")  # or groq("llama-3.3-70b-versatile"), openai("gpt-4o")
)

result = agent.run("What is the capital of France?")
print(result)

For more examples and detailed documentation, visit Docs.

Contributing

Contributions are welcome! Please read our Contributing Guide to get started. Also join Discord

License

MIT License - see LICENSE file for details