Fuz is a free software stack for human agency

Nov 6, 2026

Fuz (fuz.dev) is a free and open source software stack that I've been building full-time in public since 2019, to support simple, secure, and optimal software for the web and native systems. It has libraries, dev tools, and infra designed for self-hosters, devs, and people interested in making software who aren't devs. There's also some apps and websites on the way - these motivate everything else, but first I built the foundations to age well. The purpose is human agency: I want software one can own, control, extend, and swap out unburdened, with polished UX and technical excellence. Fuz has no investors and it will remain noncommercial, so it can focus on serving the people who use it. I lucked into an opportunity to pursue self-funded work for a substantial period of time, and after 7 years I'm announcing the results and continuation of that effort. By the end of 2027, Fuz will either become a hobby or my livelihood.

For more about my background, see my personal blog's intro post. The short version is that I've been working in startups building on open source since 2011, and I started Fuz because I love free software and its potential to improve our lives. Follow this blog's feed for more.

The Fuz stack

Fuz exists to accelerate the production of excellent software, for all kinds of purposes and platforms. Today it can ship to the web, and also native platforms with web-only GUI; mobile and native GUI are planned. The stack's third-party dependencies are opinionated to enable sharing code and patterns across projects:

  • TypeScript and Rust, with two compatible server frameworks building on Hono (TS) and Axum (Rust)
  • Svelte with SvelteKit, for UI deployed statically or dynamically, local-first or cloud, SPA or SSR
  • Postgres with embeddable support
  • Linux is primary but most tools work with macOS and Windows

Using the stack as a developer is not an all-or-nothing proposition: although it's designed for cohesion, many of its parts can be used in isolation or swapped for similar libraries. For example, fuz_css distills my 15 years of professional frontend experience and taste into a CSS framework, and although some of my frontend libraries depend on it, others make it optional (where doing so doesn't degrade integrated usage), so you could choose Tailwind and still use some parts of Fuz's frontend along with the rest of Fuz. The repos are separated by concern so you can individually use or fork them without the gravity of a monorepo. The intent is to lower the risks of adoption, streamline onboarding to the parts you care about, and keep exit straightforward.

The foundational first-party repos include:

  • fuz_app and fuz - a fullstack app library for TypeScript with a Rust twin, for building servers in either language or both if you're so inclined; the fuz repo also has low-level library crates, the fuz CLI, and fuzd daemon
  • fuz_template - starter repo for web apps with SvelteKit, native servers and tools with Rust, or both (web-only GUI for now, native in the future)
  • fuz_css - semantic-first CSS framework and design system, with flexible custom theming
  • fuz_ui - Svelte component library and frontend helpers
  • tsv - language toolchain for TypeScript/JS, CSS, and Svelte in Rust, currently ships parsers and formatters
  • gro - security-first JS runtime and toolchain on V8, the first Fuz repo, started in 2019 as a task runner that wraps Node and used every workday since. Since Deno's acquisition by Cloudflare and its announced EOL in October 2027, I'm experimenting with making Gro its own runtime that builds on Deno's foundations.
  • pglet - vibecoded embeddable postgres for going places the elephant can't fit, and faster tests; today it runs fuz_app's db tests several times faster than PostgreSQL and PGlite, but it's not an alternative to them
  • zap - convergent machine-state management: set up and ship to Linux machines from TypeScript
  • fuz-stack - an agent skill with the stack's coding conventions and patterns for TypeScript, Svelte, and Rust

For the full picture including third-party software, see fuz.dev/stack. The code is at github.com/fuzdev, and licensed with the public-domain-equivalent 0BSD. As the stack has matured, I'm starting to work on apps and websites for end-users, some in collaboration with other people. More about those later.

The AI elephant in the room

AI is reshaping the software world, and this project is driven by a person who uses AI with concern for its widespread harms to culture, livelihoods, and the environment. I built Fuz for 6 years by hand, and then in 2025 I started using agentic development tools, and I have been a heavy Claude Code user since. Looking at the list of projects above, the scope is well beyond what I imagined in 2019, but in 2026 I don't see it as absurdly over-reaching. Design is the hard and fun part to me, and AI is a big lever. I'm choosing to use AI in service of my values, which include self-determination, personal and societal and environmental health, and making high quality information and tools freely available.

For people who want software to improve our lives, AI unlocks new opportunities. I agree with those who say we are experiencing the transition to software abundance. Closed models can produce open software, and using AI doesn't imply central corporate control: open models running on private hardware can produce software that humans could not feasibly build due to cost, and costs continue to shrink. Of course, many people will continue to use exploitative software as they did before LLMs, but free alternatives are increasingly available, and quality is up to the developer. Fuz is my attempt to put abundant software to good use.

Going forward

I set out to make software under a different set of incentives than the norm, aligning directly with the people who use it, without shareholders as the supreme customer, to produce the best software I could imagine. The libraries are available for early adopters, but only a few are ready for production, and I'll share apps made with Fuz when they're ready. Fuz will not operate services beyond its community websites, due to costs and complicated incentives, simplifying its path to sustainability.

I'm seeking design collaborators to help shape the direction Fuz goes. I think the project succeeds if a community forms and deliberates through important design choices, collaboratively in public. I'm committed to working on Fuz full-time through 2027, and in 2028 I will reassess based on funding and user feedback. To contribute, no expertise is needed, only a perspective you want to share. When time permits, I hope to build a community site with the stack to supplement GitHub discussions. In the meantime you're invited to join me on GitHub to follow along or help out.

about
stack
contributing