Loading...
Loading...
Simon Willison is extending the Datasette ecosystem with new tooling that makes publishing and sharing data faster and less error-prone. He built a Claude-powered preview interface for editing datasette.io’s news.yaml, letting contributors paste YAML and instantly see how entries will render while catching Markdown or syntax issues—an example of LLMs as lightweight, repo-aware content QA tools. In parallel, he released datasette-export-database 0.3a1, enabling on-demand exports of mutable SQLite databases from Datasette. Together, the updates emphasize smoother maintenance, reproducible publishing, and more portable, auditable data workflows.
A developer launched Wordtrak, a 1v1 competitive word game inspired by Scrabble and Wordle, built as a hybrid web/mobile app using Rails and Expo. The creator designed the game loop and visuals with help from Claude (an LLM), iterated on rules, scoring, and a train-themed split-flap display UI, and used Scrabble tile scores initially. Early builds included a simple CPU opponent, randomized “Conductor” modifiers that required balancing, a lobby and stat tracking, and a curated dictionary assembled from public word lists. The project emphasizes quick matches, family-friendly play, and iterative user testing; the author leaned on LLMs for design and code assistance.
An open-source React Native app called LET (Life Events Tracker) provides an offline-first mobile and web habit, mood, and event tracker built with Expo and TypeScript. Key technologies include SQLite (expo-sqlite + Drizzle ORM) for local storage, Zustand for state management, Expo Router, NativeWind for styling, and custom SVG charts. It supports dark/light modes, week-based navigation, export/import JSON, customizable colors, swipe gestures, and works entirely offline with data kept on-device. The project was generated by prompting an AI (Claude), with the author reviewing and assembling contributions. This demo highlights AI-assisted development and offers a repo for contributors interested in mobile tracking, privacy-first apps, and cross-platform React Native tooling.
Simon Willison built a Claude-powered preview UI to simplify editing the datasette.io site’s news.yaml, which currently drives the site’s news section. He asked Claude to clone the GitHub repo, read news.yaml, and produce an artifact that previews how pasted YAML will render on the homepage while flagging Markdown or YAML errors. The tool aims to reduce friction when checking for formatting mistakes and to speed up content updates for Datasette’s changelog and news entries. This demonstrates using LLMs and Claude Artifacts for lightweight developer tooling and content QA workflows tied to a public open-source repo.
Simon Willison announced datasette-export-database 0.3a1, a new release that lets users export a copy of a mutable SQLite database on demand. The lightweight plugin/utility integrates with Datasette, offering a convenient way to produce static snapshots of live SQLite databases for sharing, archiving, or downstream processing. This matters for developers, data journalists, and operators who rely on Datasette to publish and explore datasets: on-demand exports simplify reproducible data publishing workflows, backups, and distribution without manual dump steps. The release is part of Willison’s ongoing work in the Datasette ecosystem and reinforces tooling around making live data more portable and auditable.