Favorite Code Libraries

April 20, 2026 · View on GitHub

A personal index of code libraries Daniel uses and likes — curated across languages and domains, with a short note on why each one earns its place.

Last Updated: April 2026

AI / LLM
Audio / Media
CLI & Tooling
Data / Datasets
Design / Assets
Infrastructure / DevOps
JavaScript / TypeScript
Python
Speech / ASR
Typesetting / Documents

AI / LLM

Agent Browser

Vercel Labs' Rust CLI for browser automation designed as a first-class tool for LLM agents — headless web access from agent workflows.

DSPy

Stanford framework for programming — not prompting — language models, with composable modules and optimizers replacing brittle prompt strings.

LangGraph

Graph-based orchestration library for building stateful, multi-actor LLM agents with checkpointing and human-in-the-loop support.

LlamaIndex

Python/TypeScript framework for building LLM applications over private data — retrieval, agents, and structured extraction.

OpenAI Python

Official OpenAI Python SDK — typed client covering the Responses, Chat, Embeddings, Files, and Assistants APIs.

Qdrant

High-performance vector database and similarity search engine written in Rust — filters, quantization, and cloud-native scaling for AI workloads.

Vane

Open-source AI answering engine (successor to Perplexica) — self-hostable RAG search over SearXNG with pluggable LLMs.

Audio / Media

FaceFusion

Python face-manipulation platform covering face-swap, lip-sync, and face enhancement on images and video.

Real-ESRGAN

PyTorch image/video super-resolution — practical upscaler for restoring real-world photos and anime at 4x and beyond.

yt-dlp

Feature-rich command-line downloader for YouTube and hundreds of other video/audio sites — the actively-maintained fork of youtube-dl.

CLI & Tooling

CyberChef

GCHQ's "Cyber Swiss Army Knife" — browser app for chaining encoding, encryption, compression, hashing, and parsing operations.

Homebrew

The missing package manager for macOS (and Linux) — community-maintained formulae for everyday CLI tools and apps.

PowerToys

Microsoft's open-source suite of Windows power-user utilities — FancyZones, PowerToys Run, Color Picker, and more.

rclone

"rsync for cloud storage" — sync, copy, and mount across 70+ cloud providers and protocols from the CLI.

scrcpy

Display and control Android devices from a desktop via USB or TCP/IP — no root required, low latency.

Selenium

Venerable browser automation framework and ecosystem with WebDriver bindings for Java, Python, Ruby, JS, C#, and Rust.

yadm

Dotfile manager built on top of git — track home-directory config across machines, with encryption and alt-files for per-host overrides.

Data / Datasets

Country Codes

Comprehensive dataset of country codes (ISO 3166, ITU, FIPS, and more) as CSV/JSON — a canonical lookup source for mapping between code systems.

Design / Assets

Material Design Icons

Google's official Material Design icon set — SVG, icon font, and framework-ready assets under Apache 2.0.

Infrastructure / DevOps

MinIO

High-performance S3-compatible object storage server — a drop-in replacement for AWS S3 in self-hosted and on-prem setups.

OpenAPI Specification

Canonical repository of the OpenAPI Specification — the industry-standard schema for describing HTTP APIs.

JavaScript / TypeScript

Astro

Web framework for content-driven sites — zero JS by default, with island-based hydration for interactive bits.

Better Auth

Comprehensive TypeScript authentication framework — OAuth, OIDC, SSO, 2FA, and Stripe integration out of the box.

Directus

Open-source data platform that wraps any SQL database with an instant REST/GraphQL API, admin UI, and headless CMS.

ECharts

Apache's production-ready charting and data-visualization library for the browser — rich, interactive chart types.

Expo

Framework and platform for universal React Native apps — dev, build, and deploy pipeline for iOS, Android, and web.

FingerprintJS

Open-source browser fingerprinting library for visitor identification and fraud detection — no cookies required.

Readability

Mozilla's content-extraction library — pulls the clean article body out of a web page (the engine behind Firefox Reader View).

Refine

React framework for building admin panels, dashboards, and internal B2B tools with headless hooks and data-provider adapters.

XState

Statecharts and actor-model library for JavaScript/TypeScript — model complex app logic as visualizable state machines.

Python

Apache Airflow

Workflow orchestration platform for programmatically authoring, scheduling, and monitoring data pipelines as DAGs.

Docling

IBM-led document-conversion toolkit — parses PDFs, Word, HTML, and more into structured markdown/JSON for LLM pipelines.

Marker

Datalab's fast, high-accuracy PDF → markdown/JSON converter — purpose-built for LLM ingestion pipelines.

pdfplumber

Library for detailed PDF text, table, and layout extraction — reliable where other PDF tools give up.

Scrapy

Fast, batteries-included Python framework for writing crawlers and structured web-scraping pipelines.

spaCy

Industrial-strength Python NLP library — tokenization, NER, POS tagging, and pipelines built for production workloads.

Speech / ASR

edge-tts

Python module that uses Microsoft Edge's online TTS service — high-quality neural voices without an API key.

jiwer

Fast, minimal WER/CER computation for ASR evaluation — the de-facto Python library for transcription accuracy metrics.

whisper.cpp

Plain C/C++ port of OpenAI's Whisper — runs local speech-to-text on CPU/GPU without Python.

Typesetting / Documents

Typst

Modern typesetting system — LaTeX-quality output with a saner markup language and fast incremental compile.