agentflare

# AgentFlare Python SDK 🐍

[![PyPI Version](https://img.shields.io/pypi/v/agentflare?style=flat&logo=pypi)](https://pypi.org/project/agentflare/)
[![License](https://img.shields.io/github/license/agentflare-ai/agentflare?style=flat&logo=github)](https://github.com/agentflare-ai/agentflare/blob/main/LICENSE)
[![Discord](https://img.shields.io/discord/1234567890?label=Discord&logo=discord&style=flat)](https://discord.gg/agentflare)
[![X/Twitter](https://img.shields.io/twitter/follow/agentflare?style=flat&logo=x)](https://x.com/agentflare)

The official Python SDK for AgentFlare – the infrastructure platform for AI agents. Integrate observability, deployment, and management into your Python-based AI agents with ease.

## 🚀 What is AgentFlare?

AgentFlare provides essential infrastructure for production AI agents, including real-time observability (Telepathy), cost tracking, and one-click deployments. This SDK makes it simple to enhance your agents with these features.

For more on AgentFlare, visit [agentflare.ai](https://agentflare.ai) or our [main docs](https://docs.agentflare.ai).

## ✨ Key Features

- **Telepathy Observability**: Trace decisions, monitor performance, and attribute costs in real-time.
- **Drop-in Integration**: Replace standard imports for instant enhancements.
- **Protocol Agnostic**: Works with MCP and custom protocols.
- **Analytics**: Track latency, errors, and usage.
- **Deployment Tools**: Deploy agent servers directly from your code.

## 🛠️ Installation

Install via pip:

```bash
pip install agentflare
```

Requires Python 3.10+. For development, clone the repo and install dependencies:

git clone https://github.com/agentflare-ai/agentflare.git
cd agentflare
pip install -e ".[dev]"

📚 Quickstart

Enable Observability

# Before: Standard MCP client
from mcp_sdk import Client

# After: With AgentFlare observability
from agentflare import Client  # Drop-in replacement

client = Client(api_key="your-agentflare-key")

# Now all interactions are traced automatically!
response = client.chat("Hello, AI agent!")
print(response)  # Includes reasoning, perf metrics, and costs

Advanced Usage: Custom Tracing

from agentflare import TelepathyTracer

tracer = TelepathyTracer()

with tracer.span("agent_decision"):
    # Your AI agent logic here
    decision = model.predict(input_data)
    tracer.log_cost(model_cost=0.05)  # Track spending

tracer.export()  # Send to AgentFlare dashboard

For full examples, check the examples/ directory or our quickstart guide.

📊 Documentation

• Full API Reference
• Telepathy Guide
• Deployment Tutorials

👥 Contributing

We welcome contributions! See CONTRIBUTING.md for guidelines. Start with issues labeled "good first issue".

📝 License

MIT License. See LICENSE for details.

🤝 Community & Support

• 💬 Join Discord for discussions.
• 🐦 Follow us on X/Twitter.
• 📧 Contact: hello@agentflare.ai.
• 🌟 Star this repo if it helps!

Built with ❤️ by the AgentFlare team in San Francisco.