LLM Libraries

get-md

Nano Collective

Recommended

Fast, lightweight HTML, PDF, DOCX and Markdown to Markdown converter optimised for LLM consumption. Converts files or URLs in under 100ms, with optional AI-powered conversion using a local model. Library and CLI.

• Open source
• Local-first
• Privacy-first
• Self-hostable

json-up

Nano Collective

Recommended

Fast, type-safe JSON migration tool with Zod schema validation. Fluent builder API for versioned migration chains, full TypeScript inference, async migration support and automatic version tracking.

• Open source
• Local-first
• Privacy-first
• Self-hostable

LangChain

LangChain

Framework for developing applications powered by LLMs. Provides agent orchestration, tool calling, memory and integrations with hundreds of model providers and data sources in Python and JavaScript.

• Open source
• Local-first
• Self-hostable

llama.cpp

ggml-org (Georgi Gerganov)

Recommended

LLM inference engine in C/C++ with minimal dependencies. Runs quantized GGUF models on CPU and GPU across Apple Silicon, x86 and CUDA, and powers many local AI applications.

• Open source
• Local-first
• Privacy-first
• Self-hostable

LlamaIndex

LlamaIndex

Data framework for connecting LLMs to your own data. Document loaders, indexes and query engines for building RAG pipelines and document agents in Python and TypeScript.

• Open source
• Local-first
• Self-hostable

prompt-scrubber

Nano Collective

Recommended

Local-first utility that strips identifying content — emails, secrets, file paths, URLs, phone numbers — from prompts before they reach a cloud LLM, mapping each value to a stable placeholder so responses can be rehydrated locally.

• Open source
• Local-first
• Privacy-first
• Self-hostable

Hugging Face's model-definition library for state-of-the-art transformer models in text, vision and audio. Load thousands of pretrained open models for local inference or training in PyTorch.

• Open source
• Local-first
• Privacy-first
• Self-hostable