Repository navigation
feat(members): the first mate enters center stage before it greets - #23222
Conversation
f72ab06 to
11231b5
Compare
Design Review (Opus 5.5) — ✅ PASSDesign-level review of Design-Verdict: PASS Nothing blocks. The entrance is keyed on the server's existing once-only Notes
[DESIGN-REVIEWED] 871943a Description read[DESCRIPTION-READ] 09db72af6eae012f43a2a0cd3951da4e4967057d9beb83de30231425606bfe4f That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
GPT 6.1 Review — ✅ no blocking findings (all downgraded on adjudication)GPT 6.1 flagged blocking issues on This comment is updated in place on each push. Review detailsBLOCKING -- website/src/pages/members/useFirstGreeting.ts:73 -- A delayed greet response lets the greeting appear before the entrance
Adjudication (Opus 5.5) — is blocking on each finding proportionate?F1: LOW harm, and only cosmetic. The race needs the WebSocket greeting frames to reach the client before the HTTP False positive or not applicable? A repository writer can comment: |
First Principles Review (Opus 5.5) — ✅ PASSPremise-level review of First-Principles-Verdict: PASS Nothing blocks. The one open question is whether this counts as a product-shape decision, and the opt-in Crew preview leaves that a judgement call, so it stays a note. Notes
What this change shipsInventory (10 items) — 9 justifiedIntent: the first time someone opens Mate's chat, they meet Mate as a teammate with a face before its greeting appears. This is an ADDITION.
[FIRST-PRINCIPLES-REVIEWED] 871943a Description read[DESCRIPTION-READ] bc09880994434f775a9fc65d4ea51a1f00a2ec05f90c81881fe9d527414edae7 That sha256 names the PR description this verdict read. Recompute it with |
UX Review (Opus 5.5) — ✅ PASSUX-level review of UX-Verdict: PASS Nothing blocks. The motion itself still needs a human to watch the committed GIFs: the blind reader saw only one frame of each, so it could not confirm the big face becomes the header face. Notes
[UX-REVIEWED] 871943a Description read[DESCRIPTION-READ] 79c70d65a389412bb98d103644affeb9a7134cd3f8b0c763f1a9fac633a9eb4d That sha256 names everything this verdict read: the PR description, plus the 11 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
Opus 5.5 Review — ✅ no blocking findingsReviewed Verdict parsed from the review's SHA-scoped output markers for commit False positive or not applicable? A repository writer can comment: |
11231b5 to
c99d6ff
Compare
Human judgment recorded@CrysisDeu marked the first-principles AI finding as false positive, not applicable, or explicitly accepted for
This decision applies only to this commit. A new push requires a new judgment. |
c99d6ff to
01aab93
Compare
Human judgment recorded@CrysisDeu marked the first-principles AI finding as false positive, not applicable, or explicitly accepted for
This decision applies only to this commit. A new push requires a new judgment. |
The first time Mate's chat opens and its greeting is dispatched, Mate
introduces itself before it speaks: the header name-pill's face slot (and
Mate's roster row, and its face in the folded roster's switcher chip) hold
their face unpainted; Mate's face appears large at the centre of the thread
with "Hi! I'm your first Mate." ("Hi! I'm {{name}}." once other mates
exist); after 1.5 s it flies into the pill's face slot, lands with a pop
and a ring, the slot and row fill, and the greeting turn streams in as
before.
Once per thread by construction: `useFirstGreeting` gains an `onStarted`
callback fired only for the greet request the server answers `started`,
the one outcome that means the greeting was dispatched now. A reload, a
second tab or a failed turn hear `already_greeted`, `not_empty` or an
error and show nothing; no second marker is added. Mate only: a mate
created on the New mate card already showed the person its face and name.
`prefers-reduced-motion`: faces painted from the start, the line shown
static for the same 1.5 s, no flight.
Spec: crew-mode.md "The first Mate's entrance"; rfc-crewmates-launch.md
screen 08 records the entrance as separate from the feature intro.
Closes #23207
01aab93 to
871943a
Compare
|
Human design review is required before this PR can merge. The first Mate entrance changes launch UX, while HTTP/WebSocket ordering and pop cleanup still need maintainer decisions. This comment records the design gate only; it is not a code defect. |
Problem / Motivation
Goal: The first time a person opens Mate's chat, they meet Mate -- its face enters centre stage, says hello, and flies into the header -- before Mate's greeting streams in.
Every install creates Mate (#18406) and Mate speaks first when its chat opens (
mate_welcome.py), but nothing marks the moment a person first meets Mate: the Crewmates page opens on an empty thread and a greeting appears. Someone who has never heard of crewmates sees a chat with a name in it and no sense that this is a teammate who was just handed to them. #23207 asks for the introduction of the first Mate itself, separate from the launch intro (#17154) that introduces the feature.Why it matters
The greeting introduces Mate's words, not Mate. Without an entrance, the one moment that can make "a crewmate" concrete -- a face, a line, a place it lives -- passes as a plain empty chat, for every new install.
Not a goal
startedalready means "dispatched now", so nothing is added.What changed (motivation → approach → change)
Goal: introduce Mate once, in the one tab that is about to receive its first greeting, and never again.
Approach: key the entrance on the greet request's outcome rather than on a marker of its own.
POST /api/members/{slug}/greetalready answersstartedfor exactly one request ever (the server'sO_EXCLclaim) andalready_greeted/not_empty/busy/ an error for every other, so that answer is already the once-only the issue asks for -- a reload, a second tab or a failed turn never hearstarted. This avoids a second marker and keeps the backend untouched. Mate only (isAssistantMember), since a dashboard-created crewmate's face and name were just chosen by the person.Change:
useFirstGreetinggains an optionalonStarted(slot)callback, fired only for thestartedanswer, read through a ref so a new callback identity never re-asks. It fires beforestartServerTurnso the pane is veiled before the turn's "Working" row could show.MateEntrance.tsx: the face (150px) centred on the thread column with arole="status"speech bubble ("Hi! I'm your first Mate." while Mate is the only mate; "Hi! I'm {{name}}." once others exist, the same split the kickoff makes), a 1.5 s hold, a 0.9 s flight (translate + scale of a sharp large raster, portaled above every surface likeCrewFaceFlight) to the header pill's face slot, a landing pop on the slot's own face and a ring. With no slot box (header not laid out) it cuts to landed instead of flying nowhere.prefers-reduced-motion: no face and no flight -- the bubble alone, static, for the hold.useMembersPageThreadowns the state (stage->landed-> gone), scoped to the thread it was answered for; leaving mid-entrance ends it. While the face is on stage the header pill's slot (member-pill-face), Mate's roster row face and its face in the folded roster's switcher chip hold their place unpainted (data-entrance-empty, box kept), and the chat pane is veiled, fading back as the face lands.pages.membersPage.entrance_first_mate,entrance_hello) across all catalogs withen.context.jsonnotes.docs/system-specs/modules/crew-mode.md"The first Mate's entrance";rfc-crewmates-launch.mdscreen 08 amended to record the entrance as separate from the feature intro (RFC statusaccepted, already on main).Backwards compatibility
Compatible:
useFirstGreeting's new parameter is optional and every existing call is unchanged;CrewmateSwitcher'sfaceOnStageFordefaults to''; no API, store or catalog key is renamed or removed.Removes nothing: the diff only adds a component, an optional callback, optional props and two catalog keys.
Tests
MateEntrance.test.tsx(new): opens centre stage with the large face and the line, the slot not yet landed on; flies to the slot's box and lands there (exact end transform), pops the slot face and callsonLandedthenonDone; reduced motion shows the line alone and never callsonLanded; a zero slot box cuts to landed; the host's line is what it says.useFirstGreeting.test.tsx:onStartedfires with the slot forstartedonly; stays silent onalready_greeted,not_empty,busy,not_owed,no_thread,not_crewmateand on a failed request; the latest callback hears the answer and a new identity never asks again.CrewmateSwitcher.test.tsx: the face of the mate on stage is held unpainted with its place kept; every face painted otherwise.Ran only the changed test files locally (
npx vitest runon the three), plusnpx tsc --noEmit, eslint on the touched files,I18N_BASE_REF=origin/main npm run i18n:check, the per-locale style gates and the brand gate. No Python changed.Manual verification
Recorded the real
MembersPage(thecapture/members-page.htmlentry on the Vite dev server) behind a stubbed API withwebsite/scripts/capture-first-mate-entrance.mjs, as a fresh install whose roster is Mate alone with its welcome owed and unclaimed. Each frame asserts its state before it is written: one greet request answeredstarted; the entrance on stage with the line; the pill slot, roster row and switcher face empty with their width kept; the pane veiled; landed with all faces painted; the greeting row delivered the way the WebSocket would, pane unveiled, still one greet. Three negative passes:already_greeteddraws no entrance (a reload); reduced motion shows the line alone with the faces in place and removes it after the hold.Not verified against a live gateway run of the kickoff turn; the greet outcome vocabulary is read from
mate_welcome.py(STARTED = "started").Screenshots / video
Dark, the whole entrance then the greeting:
Landed, light and dark:
More frames: light recording, centre stage, reduced motion, no replay
Media committed under
.github/screenshots/first-mate-entrance/at the request of the maintainer, referenced by commit-pinned URLs.Related Issues
Closes #23207
The maintainer waived the captain-triage wait for this issue on 2026-10-22.
Checklist
feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)Contribution License Agreement
Contributed under the project's Apache-2.0 license; see the Licensing section of CONTRIBUTING.md.