standalone packages
Packages
Sprezzature has three layers. The skills (the SKILL.md files an agent reads) stay in the main repository. The deterministic tools each graduate into their own pip-installable package, so you can use one without cloning the whole stack. A local runtime lets the tools run offline against a local model.
Each package is BSD-3-Clause, local-first, and independently versioned. They are not on PyPI yet, so install straight from GitHub.
Deterministic tools
sprezzature-figures
Figures
126 hand-authored SVG chart types, no Vega, plus the Ralph Eyeball Loop, explainability, and causal DAGs.
pip install git+https://github.com/warith-harchaoui/sprezzature-figuressprezzature-maps
Maps
Hand-authored SVG cartography split out of sprezzature-figures: an Equal Earth choropleth with an OKLCH ramp, and layered Lambert Conformal Conic situation maps, both with a real-elevation warm/cool relief.
pip install git+https://github.com/warith-harchaoui/sprezzature-mapssprezzature-colors
Colors
WCAG contrast auditing with OKLCH-neighbour fixes, colour-vision-deficiency simulation, a curated palette, and Tailwind export.
pip install git+https://github.com/warith-harchaoui/sprezzature-colorssprezzature-accessibility
Accessibility
A static HTML accessibility linter: source-decidable WCAG/WAI rules, no browser, no runtime DOM. A fast pre-commit gate.
pip install git+https://github.com/warith-harchaoui/sprezzature-accessibilitysprezzature-cli-gui
CLI → web GUI
Auto-generate a single-file HTML GUI from any Python CLI: it introspects argparse, Click, or Typer, with no build step.
pip install git+https://github.com/warith-harchaoui/sprezzature-cli-guisprezzature-ux-laws
Laws of UX
A static Laws-of-UX auditor for HTML: Hick, Miller, Fitts, Jakob, Tesler and more, decidable from source.
pip install git+https://github.com/warith-harchaoui/sprezzature-ux-lawssprezzature-audio
Audio
Local captions (whisper.cpp), speaker diarization and who-is-who, narration and translation. ML backends are optional extras.
pip install git+https://github.com/warith-harchaoui/sprezzature-audioRuntime & engine
best-engine-ai-helper
Engine picker & local runtime
Detect the machine, pick a model from a catalog, and validate a local LLM / VLM before the skills lean on it. Every skill's LLM/VLM call routes through its chat abstraction: a pluggable backend (Ollama, OpenAI-compatible, LangChain), the Ralph loop, and the writing pipeline.
pip install git+https://github.com/warith-harchaoui/best-engine-ai-helpersprezzature-local
Offline runtime
The offline runtime the skills call into: llm.py's pluggable chat backend, the Ralph Eyeball Loop, and the writing pipeline. Reads SPREZZATURE_LLM_* env vars; every other package takes it as an optional dependency, never a hard one.
pip install git+https://github.com/warith-harchaoui/sprezzature-localLooking for the agent skills instead? The SKILL.md layer ships together as the skills bundle on the Getting started section.