Skip to content

fix(members): label the post-create roster retry 'Refresh the list' - #25633

Merged
iamwhatever merged 1 commit into
mainfrom
fix/roster-retry-label-14474
Oct 29, 2026
Merged

iamwhatever merged 1 commit into
mainfrom
fix/roster-retry-label-14474

Conversation

@CrysisDeu

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: the retry on the "created, but the list didn't refresh" notice says that it re-reads the list, not that it does something to the crewmates.

After a create lands and the roster re-read fails, the Crewmates page shows Radar was created, but the list didn't refresh. with a button labelled Refresh your mates. A blind UX reader on PR #12924 was unsure whether that refreshes the crewmates themselves or only reads the roster again.

Why it matters

The button sits right after a sentence saying a crewmate was just created. A label that names the crewmates can read as "do something to Radar again" (restart it, create it again). The action only re-reads the roster, so the label should say that.

Not a goal

  • Changing what the retry does. It still re-reads the roster and, on the open path, opens the new crewmate's chat.
  • Rewording the notice sentence or other roster strings.

Intended? / Repro

Intended? No. The code comment above the button on main says the label "names the one step it repeats" so it cannot read as "create Radar again"; the step it repeats is the roster read (website/src/pages/members/MembersPage.tsx:736 retryPostCreate calls listCreated or openCreated, both of which only rosterQuery.refetch()). The label Refresh your mates names the crewmates instead of the list, which is the ambiguity the UX review found.

Repro (current main, 4efafcf): create a crewmate while GET /api/members fails after the create. The notice shows Radar was created, but the list didn't refresh. with the button Refresh your mates (website/src/pages/members/MembersPageThread.tsx:663, key pages.membersPage.create_retry_roster, website/src/i18n/locales/en.json:8064). The before frames below are that state, captured from main's catalog.

What changed (motivation → approach → change)

Symptom: a retry label that names the crewmates. Cause: the catalog string names the roster's rows, not the roster. Change: name the list, using the term the same notice and its sibling strings already use (the list didn't refresh, Refresh the list for {{name}} before adding another mate.).

  • pages.membersPage.create_retry_roster: Refresh your mates → Refresh the list, in all 12 catalogs. Each locale reuses the noun its own create_roster_failed / add_member_pending_roster strings use for the list, in the register the old label had:
    • de Liste aktualisieren, es Actualizar la lista, fr Actualiser la liste, it Aggiorna l'elenco, pt Atualizar a lista, ru Обновить список, ja 一覧を更新, ko 목록 새로고침, zh-CN 刷新列表, hi सूची रिफ़्रेश करो, bn তালিকাটি রিফ্রেশ করো.
    • en-XA regenerated with scripts/gen-pseudolocale.mjs.
  • MembersPageThread.tsx: the comment above the button quotes the new label.
  • scripts/capture-roster-retry-label.mjs (new): evidence harness for this state, desktop and mobile, dark and light. It stubs a create that lands and roster reads that fail, then presses the retry and checks it sent one more roster read, no second create and no chat message. With self it starts a Vite dev server in-process on a loopback ephemeral port.
  • scripts/capture-crewmates-page-create.mjs: its two label checks expect the new label.

Backwards compatibility

Compatible: copy only. No key, testid, API or behavior changes.

Removes nothing: no key, field or error path is removed.

Tests

  • MembersPage.newCrewmateOutcomes.test.tsx, new: the roster retry is a list re-read, not a greeting resend: its label reads the list, and a click sends no create and no message. The label is exactly Refresh the list and carries no send/again/message/greet/mates wording; a press against a still-failing roster makes another roster read and no createKirocrewAgent, memberGreet, sendChat or thread open for the new crewmate.
  • Same file: the two existing label assertions expect Refresh the list.
  • MembersPage.createLinks.test.tsx: the stay-in-chosen-chat retry test also asserts the create was sent only once.

Proof: not provable -- other: this host runs no local tests; CI runs the new tests

Manual verification

Captured with node scripts/capture-roster-retry-label.mjs self <dir> against this branch, and with main's en.json for the before frames (28/28 checks OK in each run, including the "no second create, no message" checks). Static gates run locally: eslint on the changed files, tsc -p tsconfig.app.json --noEmit, scripts/check-i18n-keys.mjs, scripts/i18n-check.mjs, gen-pseudolocale.mjs --check, scripts/check_file_line_cap.py.

Desktop, dark: before / after

before, desktop dark
after, desktop dark

Mobile, light: before / after

before, mobile light
after, mobile light

Desktop light and mobile dark

before, desktop light
after, desktop light
before, mobile dark
after, mobile dark

Related Issues

Fixes #14474

Sibling sweep

Swept: git grep -n "Refresh your mates|Refresh your crewmates|create_retry_roster"
Sibling: website/src/pages/members/MembersPageThread.tsx -- changed here (the one render site and its comment)
Sibling: website/src/pages/members/MembersPage.newCrewmateOutcomes.test.tsx -- changed here (label assertions)
Sibling: website/scripts/capture-crewmates-page-create.mjs -- changed here (label checks)
Sibling: website/src/i18n/locales/*.json -- changed here (all 12 catalogs plus en-XA)

Pattern harvest

Rule candidate: a retry button's label names the thing it re-reads (the list), never the items in it, so it cannot read as acting on those items.

Checklist

  • At most two commits (one is the norm), with a Conventional Commits title (feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)
  • Existing tests pass and new tests added for new functionality
  • Self-review completed; code follows project style guidelines
  • Documentation updated (if applicable)
  • No secrets, credentials, or internal references in the diff

Contribution License Agreement

Contributed under the project's Apache-2.0 license.

@CrysisDeu
CrysisDeu requested review from a team and KiroCrewCaptain as code owners October 29, 2026 03:58
@github-actions github-actions Bot added readiness: checking Automated validation is still running readiness: action required A blocking check or review needs attention and removed readiness: checking Automated validation is still running labels Oct 29, 2026
@github-actions

Copy link
Copy Markdown
Contributor

Design Review (Opus 5.5) — ✅ PASS

Design-level review of ebd139f4d408b47053b66bed5111f1ed9d1709fa — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS does not.

Design-Verdict: PASS

This change only edits copy. No decision record names the old label, and the PR includes before/after renders.

[DESIGN-REVIEWED] ebd139f

Description read

[DESCRIPTION-READ] 68f38f69ae1281d64df58386a769eb7efd8af5101576ac917368b00c03dccee4

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 .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions

Copy link
Copy Markdown
Contributor

Opus 5.5 Review — ✅ no blocking findings

Reviewed ebd139f4d408b47053b66bed5111f1ed9d1709fa — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] ebd139f

Verdict parsed from the review's SHA-scoped output markers for commit ebd139f4d408b47053b66bed5111f1ed9d1709fa.

False positive or not applicable? A repository writer can comment:
/ai-review override fable ebd139f4d408b47053b66bed5111f1ed9d1709fa false-positive: <reason, 20 to 500 characters>

@github-actions

Copy link
Copy Markdown
Contributor

GPT 6.1 Review — ✅ no blocking findings

GPT 6.1 completed its review of ebd139f4d408b47053b66bed5111f1ed9d1709fa and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] ebd139f

False positive or not applicable? A repository writer can comment:
/ai-review override gpt ebd139f4d408b47053b66bed5111f1ed9d1709fa false-positive: <reason, 20 to 500 characters>

@github-actions

Copy link
Copy Markdown
Contributor

UX Review (Opus 5.5) — ✅ PASS

UX-level review of ebd139f4d408b47053b66bed5111f1ed9d1709fa — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS does not.

UX-Verdict: PASS

Nothing blocks. This PR changes one label, and the blind reader read the new label correctly ("sure", "yes") in all four after screenshots.

Notes

  • This is outside what the PR is meant to fix: on the mobile shots (03/04, 07/08) no roster is visible, so the reader "wouldn't know where to look after clicking" Refresh the list. A follow-up could show the refreshed roster, or Radar's row, once the retry succeeds.

[UX-REVIEWED] ebd139f

Description read

[DESCRIPTION-READ] 16a72983203e4dfae1aba3ca48b7449a15beda90e274cd3cb81ffbd64170ceb8

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 .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions github-actions Bot added readiness: passed Eligible automated validation passed for the current revision and removed readiness: action required A blocking check or review needs attention labels Oct 29, 2026

@iamwhatever iamwhatever left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

APPROVE at ebd139f4d408b47053b66bed5111f1ed9d1709fa.

Copy-only fix: the post-create retry label becomes Refresh the list in every catalog, matching the notice's own wording, and the new test pins that the retry only re-reads the roster (no second create, greeting or message). The new capture script binds its dev server to 127.0.0.1. All checks are green and no review requests changes.

AI findings judged non-blocking

  • First Principles (PASS note; the verdict was withheld from the PR and read from the run artifact), the new capture script duplicates the existing harness: dev-only evidence tooling, no user impact.
    /**
    * Post-create roster retry label — evidence frames for the PR.
    *
    * The create of "Radar" lands (POST /api/agents answers 200) and the roster
    * re-read that follows fails (GET /api/members answers 500). The notice and
    * its retry are captured on desktop and mobile, dark and light:
    *
    * roster-retry-desktop-<theme> 1440x900, first create: the notice and the
    * retry centred in the chat column
    * roster-retry-mobile-<theme> 390x844, the same failure below md
    *
    * Each scene then presses the retry once more against the still-failing
    * roster and checks that the press issued one more roster read and no second
    * create or chat message.
    *
    * Real components against stubbed /api (Vite + Playwright), through the
    * crewmates-page-create capture entry.
    *
    * Usage (`self` starts a loopback Vite dev server for the run):
    * cd website && node scripts/capture-roster-retry-label.mjs self <outDir> [expectedLabel]
    */
    import { chromium } from 'playwright'
    import { mkdirSync } from 'node:fs'
    // `self` starts a Vite dev server in-process on a loopback ephemeral port and
    // stops it at the end; any other value is the base URL of a running one.
    const BASE_ARG = process.argv[2] || 'self'
    const OUT = process.argv[3] || '../temp-screenshots/roster-retry-label'
    const LABEL = process.argv[4] || 'Refresh the list'
    mkdirSync(OUT, { recursive: true })
    let devServer = null
    let BASE = BASE_ARG
    if (BASE_ARG === 'self') {
    const { createServer } = await import('vite')
    devServer = await createServer({ server: { host: '127.0.0.1', port: 0, strictPort: false, hmr: false } })
    await devServer.listen()
    const addr = devServer.httpServer.address()
    BASE = `http://127.0.0.1:${addr.port}`
    }
  • UX (PASS note), mobile shows no roster after the retry: outside this label fix and offered as an optional idea, not a promised follow-up.
    {/* The retry sits right after the text, not at the far edge of a
    stretched notice: the two read as one sentence, and its label
    names the one step it repeats ("Refresh the list": a re-read of
    the roster list, not a restart of any crewmate) so it cannot read
    as "create Radar again". The dismiss follows
    `postCreateDismissable`: over an empty cached roster there is

— crew-pr-reviewer (r135)

@iamwhatever
iamwhatever enabled auto-merge (squash) October 29, 2026 16:42
@iamwhatever
iamwhatever merged commit 43a8ff8 into main Oct 29, 2026
106 of 112 checks passed
@iamwhatever
iamwhatever deleted the fix/roster-retry-label-14474 branch October 29, 2026 16:44
@github-actions github-actions Bot removed the readiness: passed Eligible automated validation passed for the current revision label Oct 29, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Clarify the post-create roster retry label

2 participants