Ten weeks ago I pushed the first binary of Tabularis to GitHub. A cross-platform database client built with Tauri (Rust + React/TypeScript). One person, a late-night frustration, and a SQL editor.
Last week the repo crossed 1,000 stars. Today there are 1,086 stars, 70 forks, 15 contributors, 41 releases, and a plugin ecosystem that didn't exist eight weeks ago.
No marketing budget. No growth hacks.
I wrote a detailed post about what actually worked, what didn't, and what I'd do differently. Here are the highlights.
Ship fast, ship broken
The first week produced fifteen releases. Some were embarrassing. I shipped anyway. Nobody remembers your v0.3.0. People remember whether the tool was useful when they tried it, and whether it got better when they came back.
Build an extension point early
The decision that changed the trajectory: shipping a plugin system (language-agnostic JSON-RPC protocol) at the one-month mark. Within two weeks the community had added Redis, ClickHouse, CSV, and a Hacker News plugin. It turned users into builders.
The AI factor
I need to be honest: Tabularis would not exist without AI-assisted development. Specifically, Claude Code.
41 releases in eleven weeks, one person. That math doesn't work without a force multiplier. But AI doesn't replace experience. It amplifies it. Claude Code didn't design the plugin architecture or decide that JSON-RPC was the right protocol. Those decisions came from years of building software.
What AI did was collapse the distance between a decision and its implementation. Once I knew what to build, I could build it in hours instead of days.
The numbers at week 11
The 1,000-star milestone landed at week ten. Here's where things stand today, one week later:
| Metric | Week 4 | Week 10 | Week 11 (today) |
|---|---|---|---|
| Stars | ~270 | 1,000 | 1,086 |
| Contributors | 5 | 14 | 15 |
| Releases | 17 | 39 | 41 |
| Forks | — | 65 | 70 |
| Plugins | 1 | 7 | 7 |
| Languages | 2 | 4 | 4 |
| Issues | — | ~70 | 81 |
| Pull Requests | — | ~35 | 41 |
| Downloads | — | ~6,000 | 7,100+ |
72 countries and counting
One thing that surprised me: where those stars come from. About half of our stargazers have a public location on their GitHub profile, and they span 72 countries across every continent.
The United States and China lead, but what stands out is the long tail: South Korea, Germany, France, Brazil, Indonesia, Vietnam. Tabularis isn't a tool for one market. It's a tool for developers, and developers are everywhere.
Read the full story
The full post covers every phase week by week, what didn't work, and advice for anyone starting an open source project today.
Read the full post on tabularis.dev
If you're building something in the open, I'd love to hear your story. Drop a comment or find me on X or Discord.
Top comments (3)
This is awesome! Congrats with Tabularis, I'm checking the other posts as well
Thanks,
there’s still a lot of work to do, but we’re giving it our all 🙂
The plugin system insight is the one that scales. Once users can extend the tool themselves, you're no longer the bottleneck for every feature request — the community becomes a distributed product team. Redis and ClickHouse plugins showing up within two weeks of shipping the protocol is a strong signal you nailed the abstraction.
The "ship fast, ship broken" point is interesting in the context of Rust specifically. With most stacks, early breaking releases feel risky because bugs can be subtle and hard to trace. With Rust, v0.3.0 embarrassments tend to be loud and specific — the compiler already caught the silent ones. That changes the risk profile of shipping early: you know what broke, the compiler told you what didn't break, and v0.4.0 has a cleaner upgrade path.
One pattern I've noticed with Tauri/Rust desktop tools specifically: the "72 countries" distribution you're seeing is partly because Rust binaries ship with a predictable, minimal footprint that works on older hardware without runtime dependencies. Developers in markets with lower average hardware specs tend to notice that.
What drove the spike from ~270 stars at week 4 to 1,000 at week 10 — was there a specific community post, HN submission, or organic moment that accelerated it?