conceptual questions:
- ability to log-in to instance separate from a scene? eg, to manage scene
- single client metadata or multiple?


basic setup/infra:
- lexicon schemas
- gorm database models
- echo web server
- indexing: tap consumer, validate, filtering, tables
- hydration: skeleton-style, labels, takedowns, viewer context
- web app: pongo2, htmx, tailwind (?)
- oauth client and sessions
- label indexing: package/struct?
- dev-env with fake accounts and data

major components:
- store.Store
    => database, cache
    => manages invariants
    => high-level fetches and types
- store/models: SQL schemas
- indexer.Indexer
    => consume from tap
    => validate schemas


views:
- top bar: scrumble drop-menu; search; account/notifs drop-menu
- feed of items: user, time, interactions, tags
- scene summary bar: description, numbers, content guidelines, rules


instance admin:
- add PDS invite codes


scene service account mgmt:
- create new scenes as accounts on configured PDS (using invite codes?)
- admin endpoint to create a scene; or add/import
- instance invite codes allow anybody to create a scene
- scene admins can take actions, and instance will update scene account (records)
- lifecycle: create, migrate/import, deactivate



scene config:
- tags: any/limited/none; who
- emoji reactions: any/limited/none; who
- comments: threaded/flat/none; who
- items: types; per-type filters and limits; filter code
- membership: open/approve/invite/admin
- basic colors


tool shopping:
- CDN solution

docker-compose:
- local PDS (?)
- local PLC (?)
- postgresql (?)