Lemmy is a TypeScript monorepo designed for building AI applications, offering unified interfaces to multiple Large Language Models (LLMs) and a terminal UI framework. It targets developers building AI-powered applications who need a flexible, multi-provider LLM integration and interactive CLI experiences.

How It Works

The core of Lemmy is a unified LLM interface library that abstracts away differences between providers like Anthropic, OpenAI, and Google Gemini. It supports manual tool execution with interception, context serialization, conversation management, and streaming responses. The lemmy-tui package provides a performant terminal UI framework using differential rendering and composable components for building interactive CLI applications.

Quick Start & Requirements

• Install: git clone https://github.com/mariozechner/lemmy && cd lemmy && npm install
• Run: npm run dev (starts watch mode for all packages)
• Prerequisites: Node.js, npm.
• Resources: npm install may take a few minutes.

Highlighted Details

• Unified interface for Anthropic Claude, OpenAI, and Google Gemini.
• Type-safe tool definitions using Zod schemas.
• Differential rendering for performant terminal UIs.
• Includes example applications like claude-bridge and lemmy-chat.

Maintenance & Community

• Developed by Mario Zechner.
• No explicit community links (Discord/Slack) or roadmap provided in the README.

Licensing & Compatibility

• License: Not specified in the README.
• Compatibility: TypeScript-based, suitable for integration into Node.js applications.

Limitations & Caveats

The project is described as a "monorepo" and includes "work in progress" applications, suggesting it may still be under active development and subject to change. Licensing information is absent, which could impact commercial adoption.