FeatherAI

The lightest Agentic AI framework you'll ever see

What is FeatherAI?

FeatherAI is a lightweight Python library designed to make building AI agents incredibly simple. Whether you're creating chatbots, automation tools, or complex multi-agent systems, FeatherAI provides an elegant API that gets out of your way.

Key Features

• Simple & Intuitive API - Create powerful AI agents in just a few lines of code
• Multi-Provider Support - Works with OpenAI, Anthropic Claude, Google Gemini, and Mistral
• Tool Calling - Easily integrate custom functions and external APIs
• Structured Output - Get responses in validated Pydantic schemas
• Multimodal Support - Process text, images, and PDFs seamlessly
• Async/Await Ready - Built-in support for asynchronous execution
• Built-in Tools - Web search, code execution, and more out of the box
• Lightweight - Minimal dependencies, maximum performance

Installation

pip install feather-ai-sdk

Environment Setup

Create a .env file in your project root with your API keys:

# OpenAI (for GPT models)
OPENAI_API_KEY=your_openai_key_here

# Anthropic (for Claude models)
ANTHROPIC_API_KEY=your_anthropic_key_here

# Google (for Gemini models)
GOOGLE_API_KEY=your_google_key_here

# Mistral
MISTRAL_API_KEY=your_mistral_key_here

# For web search tools (optional)
TAVILY_API_KEY=your_tavily_key_here

Note: You only need to set the API keys for the providers you'll use.

Quick Start

Basic Agent

from feather_ai import AIAgent
from dotenv import load_dotenv

# Load environment variables
load_dotenv()

# Create an agent
agent = AIAgent(model="gpt-4")

# Run the agent
response = agent.run("What is the capital of France?")
print(response.content)  # Output: Paris

Agent with Instructions

from feather_ai import AIAgent

# Create an agent with custom instructions
agent = AIAgent(
    model="claude-haiku-4-5",
    instructions="You are a helpful assistant that provides concise answers. Always explain concepts in simple terms."
)

response = agent.run("Explain quantum computing")
print(response.content)

Agent with Tools

from feather_ai import AIAgent

# Define a custom tool
def get_weather(location: str) -> str:
    """Get the current weather for a location."""
    return f"The weather in {location} is sunny and 72°F"

# Create an agent with tools
agent = AIAgent(
    model="gpt-4",
    instructions="You are a helpful weather assistant. Use the available tools to answer questions.",
    tools=[get_weather]
)

response = agent.run("What's the weather like in San Francisco?")
print(response.content)
print(f"Tools called: {response.tool_calls}")

Structured Output

from feather_ai import AIAgent
from pydantic import BaseModel, Field

# Define your output schema
class WeatherResponse(BaseModel):
    location: str = Field(..., description="The location requested")
    temperature: int = Field(..., description="Temperature in Fahrenheit")
    conditions: str = Field(..., description="Weather conditions")
    confidence: float = Field(..., description="Confidence in answer (0-1)")

# Create agent with structured output
agent = AIAgent(
    model="gpt-4",
    output_schema=WeatherResponse
)

response = agent.run("What's the weather in Paris?")
print(response.content.location)      # Validated Pydantic object
print(response.content.temperature)   # Type-safe access
print(response.content.confidence)

Multimodal Input

from feather_ai import AIAgent, Prompt

# Create a prompt with documents
prompt = Prompt(
    text="Summarize these documents",
    documents=["report.pdf", "chart.png", "data.txt"]
)

agent = AIAgent(model="claude-sonnet-4-5")
response = agent.run(prompt)
print(response.content)

Async Execution

import asyncio
from feather_ai import AIAgent

async def main():
    agent = AIAgent(model="claude-haiku-4-5")
    response = await agent.arun("What is machine learning?")
    print(response.content)

asyncio.run(main())

Supported Models

FeatherAI supports a wide range of LLM providers:

• OpenAI: gpt-4, gpt-5-nano, gpt-4-turbo, etc.
• Anthropic: claude-sonnet-4-5, claude-haiku-4-5, claude-opus-4, etc.
• Google: gemini-2.5-flash-lite, gemini-pro, etc.
• Mistral: mistral-small-2506, mistral-large, etc.

Documentation

For detailed documentation, examples, and guides, visit our documentation site.

Topics Covered:

• Getting Started
• System Instructions
• Tool Calling
• Structured Output
• Multimodal Input
• Native Tools
• Asynchronous Execution
• Real-World Examples

Featured Projects

🍝 Piatto Cooks

An AI-powered cooking assistant that helps you discover recipes, plan meals, and get personalized cooking guidance.

• Recipe Generation
• Meal Planning
• Dietary Preferences

🎓 NexoraAI

An intelligent mentoring platform that connects mentors and mentees, providing personalized guidance and learning paths.

• Personalized Learning
• Skill Assessment
• Progress Tracking

Contributing

We welcome contributions! If you'd like to improve FeatherAI, please:

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Submit a pull request

License

FeatherAI is released under the MIT License. See LICENSE for details.

Links

• GitHub: github.com/lucabzt/feather-ai
• Documentation: lucabzt.github.io/feather-ai