Welcome to Sifaka

Sifaka is a Python library for AI-powered text improvement using research-backed critique techniques. It provides a simple API for iteratively improving text through various critique methods, with full observability and control.

Getting Started

• Installation - Setup and dependencies
• Quickstart Guide - Get running in 5 minutes
• API Reference - Complete API documentation

User Guides

• Basic Usage - Core concepts and patterns
• Critics - Available critics and their research foundations
• Validators - Input validation options
• Configuration - Customization options
• Advanced Usage - Advanced patterns and techniques

Architecture

• Architecture Overview - System design and components
• Design Decisions - Technical decisions and rationale

Why Sifaka?

Named after the Sifaka lemur known for its thoughtful movements and careful decision-making, this library helps your AI carefully consider and improve its outputs through structured critique.

Popular Use Cases

• Content Generation: Improve blog posts, articles, and creative writing
• Code Documentation: Generate and refine technical documentation
• Academic Writing: Ensure clarity and accuracy in research
• Business Communications: Polish emails, reports, and proposals
• Fact Checking: Verify claims with web search integration

Community

• GitHub Repository
• Issue Tracker
• Discussions

License

Sifaka is released under the MIT License. See the License page for details.