rbrooks/Quest-Board

Watch

Fork 0

feat(feedback-site): lane 1 of the surface mockups — the shell (#659) #669

Merged

claude-bot merged 1 commit from feat/lane1-shell-659 into main

2026-09-16 02:50:30 +00:00

claude-bot commented

2026-09-16 02:32:52 +00:00

Collaborator

The first of eight lanes, at /prototype/lane1: the furniture every other screen sits inside. It goes first because the other seven lanes all sit in it.

Six surfaces, drawn in Lapis Table

Surface	What it now shows
`SH-1` App bar	Desktop and phone, in and out of a campaign, as the GM and as a player, with and without a badge
`SH-9` Search	New (#667). Grouped and ranked results; empty / nothing-found / working / failed; the same query run by the GM and by a player
`SH-5` The bell	First drawing (#655). Unread, all-read, empty, and a deleted thread
`SH-4` Account menu	First drawing (#656). GM, player and admin
`SH-2`/`SH-3` Contents	Home and campaign, GM and player

Things the drawing is deliberately careful about

Unread is never colour alone (IS-295) — a dot and heavier text.
A deleted thread keeps its row and loses its words (IS-162). Removing the row would make yesterday's count wrong and read like a bug.
An unrecognised notification kind is dropped, not drawn wrong (IS-048).
GM-only search results are filtered, not hidden. The player's panel is built from a list the GM-only records never entered, and the GM's copy marks them so Dana can see what her players cannot.
Theme is three choices, not a toggle. A two-state switch cannot say "follow the system", which is the default.

Pronouns (#653)

The sample now carries a player's pronouns separately from their character's, and they deliberately differ: Priya (she/her) plays a he/him rogue, Marcus (he/him) plays a she/her druid. My first draft showed Priya her character's pronouns, which is exactly the error #653 exists to prevent — so the fiction now makes the two impossible to conflate, and search renders both: "Wren Halloway · Druid 6 · she/her · played by Marcus (he/him)".

Four questions drawn both ways rather than decided

The phone bar (wordmark vs a real search box), grouped vs ranked results, whether there is a footer at all, and where the account menu lives on a phone. Each is drawn, not described, so the owner can choose by looking.

The shell moves to shared files

shared/shell.css and shared/shell.js (window.QBS), so the remaining seven lanes share one copy instead of each carrying its own. look.html keeps its inline copy untouched — it is the frozen record of the look decision, and a change made here must not be able to break it. The two are allowed to diverge; the header comment says so.

Verified by rendering in Chromium at 1500px, light and dark: no console output, no page errors. Three faults in my own first draft were found that way and fixed — a caption promising a search box the drawing did not have, a caption pointing at the wrong row, and desktop bars squeezed into 340px columns where the breadcrumb wrapped at a width no desktop has.

Tests: 77 passed (7 new for the routes).

🤖 Generated with Claude Code

The first of eight lanes, at `/prototype/lane1`: the furniture every other screen sits inside. It goes first because the other seven lanes all sit in it. ## Six surfaces, drawn in Lapis Table | Surface | What it now shows | |---|---| | `SH-1` App bar | Desktop and phone, in and out of a campaign, as the GM and as a player, with and without a badge | | `SH-9` Search | **New** (#667). Grouped and ranked results; empty / nothing-found / working / failed; the same query run by the GM and by a player | | `SH-5` The bell | **First drawing** (#655). Unread, all-read, empty, and a deleted thread | | `SH-4` Account menu | **First drawing** (#656). GM, player and admin | | `SH-2`/`SH-3` Contents | Home and campaign, GM and player | ## Things the drawing is deliberately careful about - **Unread is never colour alone** (`IS-295`) — a dot *and* heavier text. - **A deleted thread keeps its row and loses its words** (`IS-162`). Removing the row would make yesterday's count wrong and read like a bug. - **An unrecognised notification kind is dropped, not drawn wrong** (`IS-048`). - **GM-only search results are filtered, not hidden.** The player's panel is built from a list the GM-only records never entered, and the GM's copy marks them so Dana can see what her players cannot. - **Theme is three choices, not a toggle.** A two-state switch cannot say "follow the system", which is the default. ## Pronouns (#653) The sample now carries a **player's** pronouns separately from their **character's**, and they deliberately differ: Priya (she/her) plays a he/him rogue, Marcus (he/him) plays a she/her druid. My first draft showed Priya her character's pronouns, which is exactly the error #653 exists to prevent — so the fiction now makes the two impossible to conflate, and search renders both: *"Wren Halloway · Druid 6 · she/her · played by Marcus (he/him)"*. ## Four questions drawn both ways rather than decided The phone bar (wordmark vs a real search box), grouped vs ranked results, whether there is a footer at all, and where the account menu lives on a phone. Each is drawn, not described, so the owner can choose by looking. ## The shell moves to shared files `shared/shell.css` and `shared/shell.js` (`window.QBS`), so the remaining seven lanes share one copy instead of each carrying its own. **`look.html` keeps its inline copy untouched** — it is the frozen record of the look decision, and a change made here must not be able to break it. The two are allowed to diverge; the header comment says so. Verified by rendering in Chromium at 1500px, light and dark: no console output, no page errors. Three faults in my own first draft were found that way and fixed — a caption promising a search box the drawing did not have, a caption pointing at the wrong row, and desktop bars squeezed into 340px columns where the breadcrumb wrapped at a width no desktop has. Tests: 77 passed (7 new for the routes). 🤖 Generated with [Claude Code](https://claude.com/claude-code)

claude-bot added 1 commit

2026-09-16 02:32:53 +00:00

feat(feedback-site): lane 1 of the surface mockups — the shell (#659 )

CI / Summarisation accuracy eval harness (stub provider) (pull_request) Successful in 53s

Details

CI / Docker image build (pull_request) Successful in 30s

Details

CI / Bot/backend version sync (pull_request) Successful in 41s

Details

CI / Backend lint (ruff) (pull_request) Successful in 44s

Details

CI / Frontend tests, audit, and build (pull_request) Successful in 2m51s

Details

CI / Bot tests and audit (pull_request) Successful in 2m11s

Details

CI / Backend migration, tests, and audit (pull_request) Successful in 9m1s

Details

CI / Synthetic session harness (no GPU, no LLM) (pull_request) Successful in 16m35s

Details

774eb86874

The first of eight lanes: the furniture every other screen sits inside.
Nothing here is a page in its own right, which is why it goes first —
the other seven lanes all sit in it.

Six surfaces drawn, in Lapis Table:

- SH-1, the app bar, on desktop and phone, in and out of a campaign,
  as the GM and as a player, with and without a badge.
- SH-9, campaign search (#667, new): grouped and ranked results, the
  empty/none/working/failed states, and the same query run by the GM
  and by a player so the audience filter is visible rather than
  asserted.
- SH-5, the bell (#655, first drawing): unread marked by a dot AND by
  weight, never by colour alone (IS-295); a deleted thread keeps its
  row and loses its words (IS-162); an unknown kind is dropped rather
  than drawn wrong (IS-048).
- SH-4, the account menu (#656, first drawing): name, pronouns,
  Profile, Help, Admin for admins, theme as three choices rather than
  a toggle, sign out.
- SH-2 and SH-3, the Contents, on Home and inside a campaign, GM and
  player. Carries the three entries #656 accepted: Discussions,
  Members & characters, Table rules & safety.

Four questions are drawn both ways rather than decided: the phone bar,
grouped vs ranked results, whether there is a footer at all, and where
the account menu lives on a phone.

The shell itself moves to shared/shell.css and shared/shell.js, so the
remaining seven lanes share one copy instead of each carrying its own.
look.html keeps its inline copy untouched: it is the frozen record of
the look decision, and a change here must not be able to break it.

Pronouns (#653): the sample now carries a player's pronouns separately
from their character's, and they deliberately differ — Priya (she/her)
plays a he/him rogue, Marcus (he/him) plays a she/her druid. Showing
one where you meant the other gets a real person wrong, so every
surface that names someone has to say which it is using.

Tests: 77 passed (7 new for the routes).

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

claude-bot scheduled this pull request to auto merge when all checks succeed

2026-09-16 02:32:56 +00:00

claude-bot referenced this pull request

2026-09-16 02:49:13 +00:00

[UX] Mock up every catalogued surface, for players and GMs, before building #659

claude-bot referenced this pull request

2026-09-16 02:49:28 +00:00

[Backend+Frontend] Pronouns for players, and pronouns shown wherever a person or character is named #653

claude-bot merged commit 4e4e1c954d into main

2026-09-16 02:50:30 +00:00

claude-bot deleted branch feat/lane1-shell-659

2026-09-16 02:50:30 +00:00