AudioChart is a voice-interface PWA that reads nautical chart data — bearings, hazards, buoys — aloud, and keeps working fully offline once charts are downloaded. The code is public; here's how it's put together.
The repo is organized as a pipeline: raw chart data is prepared once, served to clients, and consumed by an installable offline web app.
Prepares raw nautical chart data — buoys, hazards, depths, coastline — into the format the app consumes.
Backend service that fronts the prepared chart data for the client.
The installable PWA — map, voice output, offline chart caching, and the on-screen controls.
CI that builds and deploys the app to GitHub Pages on every push to main.
The repo keeps its own history and design decisions in plain markdown.
Design specs for features, including the region pipeline and app structure.
CHANGELOG.mdPer-commit through mid-May 2026, then condensed multi-week summaries from late July onward — see its own header note.
RECENT_CHANGES.mdA frozen one-time write-up on a single related feature set (Focus Target, Simulate Heading) — not a rolling log; see CHANGELOG.md for what's shipped since.
ADVENTURE_TESTS.mdReal-world GPX test routes used to validate navigation and voice output.
COMMITS.mdA running log of development history alongside the chart data itself.
test/Automated tests backing the chart pipeline and app behavior.
The fastest way in is the repo itself — SPEC.md covers the architecture in more depth than any landing page will.
AudioChart's commit history is a mix of human and AI-authored changes — a live look at what iterating on a real app with Claude Code looks like over hundreds of commits.
Every feature, fix, and chart region added to AudioChart is tracked in the open — from the original offline-PWA groundwork to the latest chart data fixes. Browse the commit log to see the app's actual development process, not just the finished product.