forge-kaif 0.4.0

A modular Python infrastructure runtime for config, logging, retries, AI, and more

forge

The developer runtime for AI-era Python.

---

forge is a modular Python runtime that eliminates the undifferentiated heavy lifting of backend development — config, logging, retries, AI model integration, health checks, caching, and more.

from forge.ai import complete, Message

response = await complete(
    messages=[Message.user("What is 2+2?")],
    model="gpt-4o",
)
print(response.content)

Features

forge is not a web framework. It is the infrastructure layer that sits below your web framework — the reusable runtime that every Python backend service needs.

• Config — typed, validated, layered configuration from env vars, .env, and TOML
• Logging — structured JSON logging with automatic trace context propagation
• Retry — exponential backoff, jitter, circuit breaker, and timeout management
• AI — unified interface across OpenAI, Anthropic, Gemini, and Ollama
• Health — Kubernetes-compatible /health and /ready endpoints
• Cache — decorator-based caching with in-memory and Redis backends
• Validation — Pydantic-integrated input validation with consistent error responses
• CLI — forge init, forge check config, forge add module

Installation

pip install forge-kaif

With optional extras:

pip install "forge-kaif[openai]"       # OpenAI provider
pip install "forge-kaif[all]"           # all optional providers
pip install "forge-kaif[dev]"           # development tooling

Quick Start

import asyncio
from forge.ai import complete, Message

async def main():
    response = await complete(
        messages=[Message.user("Hello, world")],
    )
    print(response.content)

asyncio.run(main())

Documentation

Full documentation is available at ForgeKaif.

Contributing

We welcome contributions! See CONTRIBUTING.md to get started.

License

MIT — see LICENSE for details.