Maid

MapLibre iD Editor: an iD-inspired OpenStreetMap editor to build your own single-purpose editor on.

Try it: https://osm-editor-kit.github.io/maid/ (OpenStreetMap, or its development server for testing: switch in the account menu)

[!NOTE] Claude Opus 5.5 wrote and reviewed the code of this project. @tordans specified what it should do and tested and verified its user-facing behavior.

OpenStreetMap data gets more detailed by the minute, and communities form around their topics: road and bike infrastructure, street parking, playgrounds, climbing routes. Each deserves an editor made for its topic and limited to it.

One central Maid, many editors

This repository is the central Maid. An editor is an app in apps/ with a config that picks modules, presets and filters; everything a topic needs beyond that is a module in packages/. We build our own editors the same way: apps/maid (general), apps/maid-bicycle (road and bike infrastructure), apps/maid-parking (street parking) and apps/maid-indoor (indoor and building levels).

To make your own editor:

1. Fork this repository.
2. Add your app and modules next to ours, for example apps/maid-golf and packages/module-golf. ARCHITECTURE.md shows how, step by step. Coding agents work well here: AGENTS.md gives them the rules, the scripts and the checks.
3. Deploy: set MAID_APP=maid-golf in deploy.env (or as a repository variable) and push; the GitHub Action publishes that app to GitHub Pages. See docs/deploying.md.
4. Stay in sync: merge this repository into your fork now and then. Your work lives in your own app and module folders, so the merge has nothing to argue about. Where the core lacks something, change it in a small commit of its own and send it as a pull request, so your fork and the central Maid stay the same there.

What it is made of

Kept from iD: the tagging schema and presets, the imagery index, the look of the data and the editing behaviour mappers know. Rewritten: rendering (MapLibre GL through react-map-gl), the interface (React, TypeScript, Tailwind) and the tooling (Bun, Vite). The data model follows Rapid. Maid is faster than both: the map is ready, drawn and fully loaded sooner in every benchmark scene (numbers).

Develop

bun install
bun run dev        # the four apps and the local mock API
bun run check      # types, lint, format, texts, unit tests
bun run check-all  # the browser checks

Everything runs offline against frozen fixtures; automated runs send no request to OSM servers. bun run refs-setup and bun run compare show iD, Rapid and Maid side by side, and bun run bench measures them.

More: ARCHITECTURE.md (how it is built), AGENTS.md (how to work here), docs/build-out/steps.md (what is done and next), docs/architecture/, docs/adr/, docs/research/.