This CLI tool, create-llama, simplifies the creation of LlamaIndex applications by providing pre-configured templates and a guided setup process. It targets developers looking to quickly build AI-powered applications, offering a choice between a full-stack Next.js frontend with a TypeScript backend or a separate Next.js frontend connecting to a Python FastAPI backend.

How It Works

The tool leverages a CLI-driven interactive or non-interactive setup to generate a project structure. Users can select from pre-defined use cases like Agentic RAG or Data Analysis. The generated application includes a Next.js frontend powered by shadcn/ui components, acting as a chat interface. Backends can be Next.js (using LlamaIndex.TS) or Python FastAPI (using the llama-index Python package), both offering streaming chat and file upload endpoints. It defaults to OpenAI models but allows customization.

Quick Start & Requirements

• Install/Run: npx create-llama@latest
• Prerequisites: Node.js, npm/yarn/pnpm. OpenAI API key recommended.
• Setup: Interactive setup is quick (~1 sec to generate code). Generating code and installing dependencies takes ~2 min.
• Docs: LlamaIndex TS/JS docs, Python docs

Highlighted Details

• Offers two backend choices: Next.js (LlamaIndex.TS) or Python FastAPI (llama-index Python package).
• Supports ingesting various file types (PDF, TXT, CSV, MD, DOCX, HTML) with Next.js/Express, and more (video, audio) with Python.
• "Pro mode" allows granular customization of vector stores, tools (Code Interpreter, Web Search), data sources, and observability.
• Includes pre-configured use cases like Agentic RAG, Data Analysis, and Report Generation.

Maintenance & Community

• Inspired by and adapted from create-next-app.
• Links to LlamaIndex documentation are provided.

Licensing & Compatibility

• The README does not explicitly state the license for create-llama itself. The underlying LlamaIndex libraries have various licenses (e.g., MIT for some Python components, Apache 2.0 for others). Compatibility for commercial use depends on the specific licenses of the chosen backend and dependencies.

Limitations & Caveats

The tool defaults to OpenAI models, requiring an API key for full functionality, though other LLMs are supported via manual code edits. Indexing new data requires re-running the generation command.