Skip to content

chore(i18n): rename Crew Members to Agents across the dashboard - #11869

Closed
kyleseaman wants to merge 1 commit into
mainfrom
chore/agents-vocabulary
Closed

kyleseaman wants to merge 1 commit into
mainfrom
chore/agents-vocabulary

Conversation

@kyleseaman

@kyleseaman kyleseaman commented Sep 18, 2026 •

Copy link
Copy Markdown
Contributor

Problem / Motivation

The dashboard had three names for one thing. The left rail said Crew Members. The page it opened called every row a member — "Add member", "Search members…", "No crew members yet." And three of that page's own filters described their contents as Crews: "Crews that ship with Kiro Crew".

They are all the same object: a standing agent you hold a thread with.

Why it matters

A reader cannot tell whether a "crew member", an "agent" and a "crew" are three things or one. That doubt is expensive in the surface where it lands: the rail row, the roster, the create form, and the switch in Settings that reveals the page. Issue #2622 recorded the same split one page over, on the Agents tab, and was closed by renaming only that tab — this page kept its old vocabulary.

What changed (motivation → approach → change)

Everything a user reads on this surface now says Agents, from the rail row down to the create form's last field hint. The switch in Settings → Developer → Feature Previews that reveals the page is Agent Mode — it names a mode rather than the page, because that is what a preview toggle is.

Where Before After
Nav rail, New menu, page title 🟥 Crew Members 🟩 Agents
Roster controls 🟥 Add member · Search members… · No crew members yet. 🟩 Add agent · Search agents… · No agents yet.
Roster count 🟥 {{count}} members 🟩 {{count}} agents
Side panel's first tab 🟥 Crew summary 🟩 Agent summary
Source filters 🟥 Crews that ship with Kiro Crew 🟩 Agents that ship with Kiro Crew
Drawer notices 🟥 Nothing wakes this member automatically. 🟩 Nothing wakes this agent automatically.
Editor link 🟥 Open crew manager · Edit in crew manager 🟩 Open agent editor · Edit this agent
Create form body 🟥 What this member uses · hand work to this member 🟩 What this agent uses · hand work to this agent
Settings switch 🟥 Crew Members 🟨 Agent Mode
Route, filenames, catalog keys 🟦 /members, MembersPage.tsx, membersPage.* 🟦 unchanged

🟩 added · 🟨 changed · 🟥 removed · 🟦 unchanged

The create form carries two parallel vocabularies — *_member keys for the roster origin, unsuffixed ones for the Capabilities origin — so renaming only the dialog title would have left the form calling an agent a member one field in, which CrewRoster.test.tsx exists to forbid. The roster origin's fourteen strings move with the title, in every catalog. Three of those _member values are now identical to their unsuffixed siblings, because carrying the old noun was the only thing they ever did; collapsing them removes a key from thirteen catalogs and is left to a follow-up.

Only catalog VALUES moved. No key was renamed, so no call site, no parity gate and no dead-key gate moves with it, and the eleven translated catalogs under website/src/i18n/locales/ carry the same rename — each reusing the word for "agent" its own nav.agent_capabilities already used, and the word for "mode" its own focus_mode already used. en-XA.json is regenerated, as is the translator context in website/src/i18n/en.context.json. Russian's member_count_few and _many have no English counterpart, so they are renamed too; left alone, one key family would have shipped "1 агент / 2 участника". The CJK roster counters move off people-counters (名 · 명 · 位) onto object-counters, because the noun is software now.

website/src/surfaces/builtins.tsx holds the surface's English label fallback, so it moves with the catalog. The settings registry derives a card's id from its label, so the Feature Previews card id becomes developer.agent-mode in the regenerated website/src/components/commandPalette/settingsRegistry.gen.ts and src/kiro_crew/docs/settings-registry.generated.json; website/src/hooks/useSettingHighlight.ts points at the new id and keeps every older spelling in its legacy map.

Backwards compatibility

Compatible: no catalog key, API field, route or config value changes, so nothing that reads this data stops working. The one renamed identifier is the settings-registry card id (developer.crew-members → developer.agent-mode), which is derived from the label and only ever addresses a highlight; resolveLegacyHighlightId maps that id and the older developer.crew-members-and-crew-mode and developer.agents onto the new one, so a saved link or palette entry still lands on the switch.

Tests

website/src/pages/members/, website/src/test/ and website/playwright/ carry the assertions that pin this copy: the roster's empty state, counts, search, sort-and-filter label, summary-tab accessible name, edit affordances, memory-binding notices, the create dialog's name and body, the private-memory guidance, plus the preview-switch name and the surface label. Each one asserts the shipped string, so a partial rename fails rather than passing quietly. The form-coherence test in CrewRoster.test.tsx still asserts that the dialog title and its body use ONE noun — now "agent" — and the edit-later-note test now asserts both origins read the same, which is what the collapse of the two vocabularies means.

website/scripts/capture-agent-mode-card.mjs is new. It shoots the Settings card gateway-free and refuses to write a PNG unless the rendered text starts with the expected label, names the Agents page, and contains no "Crew Members" — so a stale frame cannot be attached as evidence for a rename that did not land.

Manual verification

Both frames below are real renders, not mockups: the roster comes from the existing capture-members-page.mjs entry mounting the real page, the card from the new gated capture against the built SPA. Reading them is what caught two strings a text search had missed — the drawer's memory-binding notice and its "Open crew manager" button.

Screenshots / video

The Agents page — rail row, roster, counts, search, the "Agent summary" tab, and the drawer's renamed notice and button:

Agents page with a roster of four agents and the Agent summary tab open

Settings → Developer → Feature Previews:

The Agent Mode switch, described as revealing the Agents page

The create-agent dialog, reached from the roster's add entry. Added for the UX review, which could not evaluate this surface from the earlier frames:

The create-agent dialog: Add agent title, What this agent uses, and the Create agent button

The sort-and-filter popover open on the roster. Its source descriptions are hover title attributes, so the capture proves the surface while the script asserts the exact renamed copy:

The Sort and filter agents popover, open, showing the Mine / Built-in / From packages origin rows

Related Issues

no linked issue: #2622 covered the same split on the Agents tab and is already closed; this is the sibling surface, tracked by nothing open.

Checklist

  • At most two commits (one is the norm), with a Conventional Commits title (feat|fix|docs|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

@kyleseaman
kyleseaman requested review from a team and cixuuz September 18, 2026 19:07
@kyleseaman

Copy link
Copy Markdown
Contributor Author

Intent: Make the dashboard call a standing agent by one name. The rail, the page it opens, that page's create form and the Settings switch that reveals it all said different things — "Crew Members", "member", "Crews" — for the same object; every user-visible string on that surface now says "Agents", and the preview switch says "Agent Mode".
Not a goal: Renaming anything a user cannot read — the /members route, MembersPage.tsx, the membersPage.* catalog keys, member_id, or the backend's own vocabulary. Also out of scope: the memory panels (memoryV2.*) and the Capabilities inheritance pane, which still say "member" for the same object and are a separate surface; and collapsing the three *_member catalog values that this rename made identical to their siblings.

@github-actions github-actions Bot added the readiness: checking Automated validation is still running label Sep 18, 2026
@github-actions

github-actions Bot commented Sep 18, 2026 •

Copy link
Copy Markdown
Contributor

GPT 5.6 Review — ✅ no blocking findings

GPT 5.6 completed its review of d4cdc03b0af4a8f5113f82e7550dd76d28067dda and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] d4cdc03

False positive or not applicable? A repository writer can comment:
/ai-review override gpt d4cdc03b0af4a8f5113f82e7550dd76d28067dda: <one-sentence reason>

@github-actions

github-actions Bot commented Sep 18, 2026 •

Copy link
Copy Markdown
Contributor

Design Review (Fable 5) — ✅ PASS

Design-level review of d4cdc03b0af4a8f5113f82e7550dd76d28067dda — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

Design-Verdict: PASS

Verify the untouched "New members get private Memory V2. Existing members keep…" notice is intentional — it renders beside strings this PR renamed on the same page.

Suggestions

  • Rename pages.kiroCrewAgentsPage.bindings_preview_notice in the same pass: it is the last user-visible "member" on a surface whose every neighbouring string now says "agent", and CrewRoster.test.tsx:281 already pins it.

[DESIGN-REVIEWED] d4cdc03

@github-actions

github-actions Bot commented Sep 18, 2026 •

Copy link
Copy Markdown
Contributor

First Principles Review (Fable 5) — 🟡 CONCERNS

Premise-level review of d4cdc03b0af4a8f5113f82e7550dd76d28067dda — why this exists and whether the shipped surface is the smallest honest version. Updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

First-Principles-Verdict: CONCERNS

"Agent Mode" coins a switch-only name by preference, and two "member" strings survive on the surfaces the description declares fully renamed.

Not justified as shipped

  • Item 3 — symptom-level (point patch): "Everything a user reads on this surface now says Agents" — yet grepping : ".*\bmembers?\b in en.json leaves 2 crew-noun siblings on these same surfaces: pages.kiroCrewAgentsPage.bindings_preview_notice ("New members get private Memory V2…", rendered at website/src/pages/KiroCrewAgentsPage.tsx:1940) and components.jobForm.member_pinned_hint ("Bound to this member…", website/src/components/JobForm.tsx:38, whose own comment says "A member surface says 'member' throughout").
    Clears when: both values are renamed across the thirteen catalogs, or the description names them as the remainder.
  • Item 6 — inherited: the "Mode" coinage rests on "that is what a preview toggle is", while the pin it rewrites ("the label names the page the flag holds") would now yield the collision-free "Agents" — the bare-"Crew" clash the new comment blames on that rule came from the old page name, not the rule.
    Subtraction: relabel the switch "Agents" (the page's own name), deleting the third name form and the pin rewrite.
    Clears when: a named report of a page-named switch failing someone, or a maintainer keeps the mode naming.
  • Item 9 — undeclared: the sessions notice rewording ("it belongs to {{surface}}") is absent from the description's table; grammar-derived, no premise risk.
  • Item 10 — rides along: the 326-line capture script is evidence tooling beside the fix; it follows the repo's convention (582 capture-*.mjs siblings counted), no premise risk.

What this change ships

Inventory (10 items) — 6 justified

Intent: make every name a user reads for a standing agent say "Agents", ending the member/crew/agent three-way split (#2622's sibling surface) — a FIX.

  1. Nav rail row, New menu entry, and page title now read "Agents" — justified
  2. Roster controls, count, search, and empty state say "agent" — justified
  3. Drawer notices, "Agent summary" tab, "Open agent editor" renamed — symptom-level: 2 counted "member" strings remain on these surfaces
  4. Filter source descriptions say "Agents that…" instead of "Crews that…" — justified
  5. Create form speaks one noun ("agent") from both origins; coherence pin kept — justified
  6. Settings preview switch relabeled "Crew Members" → "Agent Mode" — inherited: mode coinage over the page's name is a stated preference
  7. Settings card id becomes developer.agent-mode; old ids still resolve — justified
  8. Eleven locales, en-XA, and translator context move, incl. Russian plurals and CJK counters — justified
  9. Sessions notice reworded to "it belongs to {{surface}}" — undeclared: not in the description's table
  10. New gated screenshot script committed in-tree — rides along: PR evidence, not the fix

[FIRST-PRINCIPLES-REVIEWED] d4cdc03

@github-actions

github-actions Bot commented Sep 18, 2026 •

Copy link
Copy Markdown
Contributor

Opus 4.8 Review — ✅ no blocking findings

Reviewed d4cdc03b0af4a8f5113f82e7550dd76d28067dda — this comment is updated in place on each push.

Review details

The change is a value-only i18n rename. The one behavioral surface — the deep-link plumbing — is internally consistent: registry id developer.agent-mode with labelKey rendering "Agent Mode" matches the DOM data-setting-label, both legacy ids map onto the new id, and the nav surface persists by the unchanged navId (members), not by the label. No catalog keys were renamed, so parity/dead-key gates and call sites don't move. Nothing survives falsification, and my own scan of the code-behavior files surfaces no grounded defect.

No findings.

[OPUS-REVIEWED] d4cdc03

Verdict parsed from the review's SHA-scoped output markers for commit d4cdc03b0af4a8f5113f82e7550dd76d28067dda.

False positive or not applicable? A repository writer can comment:
/ai-review override fable d4cdc03b0af4a8f5113f82e7550dd76d28067dda: <one-sentence reason>

@github-actions

github-actions Bot commented Sep 18, 2026 •

Copy link
Copy Markdown
Contributor

UX Review (Fable 5) — 🟡 CONCERNS

UX-level review of d4cdc03b0af4a8f5113f82e7550dd76d28067dda — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

UX-Verdict: CONCERNS

The rename leaks: "Open agent editor" lands on an always-visible notice still calling agents "members" twice, and the avatar builder still says "crew editor."

Watch

  • kiroCrewAgentsPage.bindings_preview_notice — "New members get private Memory V2. Existing members keep their configured Memory V1…" renders unconditionally on the agent editor (website/src/pages/KiroCrewAgentsPage.tsx:1940), ten lines below editor_options_load_failed, which this PR did rename. Every visit to the rename's own escape-hatch destination re-poses the member-vs-agent doubt the PR exists to remove (every visit × comprehension friction × permanent). Fix: rename this value across the thirteen catalogs like its siblings.
  • avatarBuilder.apply_hint (en.manual.json:1432) — "press 'Save changes' in the crew editor" names a surface no label calls "crew editor" anymore, so the instruction points at nothing the user can find. Fix: "in the agent editor".

Suggestions

  • en.manual.json open_crew_members_desc: "each agent is standing, with its own thread" reads as posture; use the sibling crew_members_desc shape — "every agent you have, each with its own thread".

[UX-REVIEWED] d4cdc03

@github-actions github-actions Bot added readiness: action required A blocking check or review needs attention and removed readiness: checking Automated validation is still running labels Sep 18, 2026
The product had three nouns for one thing. The rail said "Crew Members",
the page it opened called every row a "member", and three of its filters
said "Crews". All of them are the standing agents you hold a thread with,
so all of them now say "Agents".

The nav rail, the sidebar create-menu entry and the page title are
"Agents". The page's own vocabulary follows: "Add agent", "Search agents",
"No agents yet.", "{{count}} agents", the four status-filter descriptions,
the side-panel "Agent summary" tab and every error notice. The Settings >
Developer > Feature Previews switch that reveals the page is "Agent Mode".

Values only: no catalog KEY was renamed, so no call site, parity gate or
dead-key gate moves. All eleven target catalogs carry the same rename,
each reusing the rendering of "agent" already in its own
nav.agent_capabilities. The pseudolocale and both settings-registry
artifacts are regenerated.

The registry derives a card id from its label, so the Feature Previews
card id becomes developer.agent-mode. Both earlier spellings join
LEGACY_ID_EXACT so a saved deep link still lands on the switch.

The route, the filenames and the membersPage.* catalog keys keep the
"member" spelling; none of it is user-visible.
@kyleseaman
kyleseaman force-pushed the chore/agents-vocabulary branch 2 times, most recently from 01d4455 to d4cdc03 Compare September 18, 2026 21:51
@kyleseaman

Copy link
Copy Markdown
Contributor Author

FIXED — both evidence gaps closed, and the Watch item was real.

Cannot-evaluate #1 (create-agent dialog) and #2 (sort-and-filter popover): both
surfaces now have committed frames, captured by an assertion-gated script that
refuses to write a PNG unless the required strings render, and that deletes each
target before driving the page so a failed run cannot leave a previous frame
behind as current evidence.

Watch (five missed strings): correct on all five, and they were missed because
they sit on edge paths that neither a grep for the old noun nor a screenshot of
the happy path reaches. Renamed in English and in all 11 locale catalogs:
member_already_exists, bindings_preview_info, editor_options_load_failed,
discard_also_crew_edits, noOtherMembers.

Two grammar defects surfaced while fixing them, both caused by the rename and
neither caught by any gate:

  • member_already_exists needed its article corrected: "A member named" ->
    "An agent named".
  • discard_also_crew_edits carries a pronoun that points at the schedule, not
    at the agent. In es/fr/it the original pronoun deliberately DISAGREED in gender
    with the crew noun (ella/elle/essa against a masculine noun) because the
    schedule is feminine in those locales, and that disagreement was the only thing
    identifying the referent. Making the possessor masculine re-aimed the pronoun at
    the agent, so a dialog about discarding a schedule read as deleting the agent.
    All four Romance locales plus ru now name the referent explicitly instead of
    pronominalizing it. de was already safe via invariant damit.

@kyleseaman

Copy link
Copy Markdown
Contributor Author

ACCEPT AND DEFER — the finding is correct; the fix belongs outside this PR.

"Open agent editor" does navigate to CREW_MANAGER_PATH (/capabilities?tab=crews),
whose tab label stays "Crews". The cross-surface split is real and this PR does
narrow it rather than close it.

Deferred because the destination is the Capabilities surface, which this PR
declares a non-goal in its body: renaming that tab pulls in the memoryV2.*
panes and the Capabilities inheritance pane (~35 more user-visible strings) on a
surface with its own reviewers. Independent evidence that the boundary is where
the concept actually changes: crewCapabilities.discardDraftBody on that pane
still reads "No other member is affected.", the exact sibling of the string this
PR renamed on the Agents page.

Filing a linked follow-up for the Capabilities vocabulary rather than widening a
copy change into a second surface mid-review.

@kyleseaman

Copy link
Copy Markdown
Contributor Author

ACCEPTED AS AN OWNER DECISION — not reversed, but the stale rationale it cited
was a real defect and is fixed.

The label is the repository owner's explicit instruction: the page is "Agents"
and the Dev toggle is "Agent Mode". I am not substituting my own judgement for
that, so the subtraction is declined.

The lane is right that the change left a contradiction behind, though. The pin at
website/src/test/previewSurfaces.test.tsx had its ASSERTION updated to
/^agent mode$/i while its comment still claimed "The label names the page the
flag holds" -- a rationale this change deliberately reverses. That comment now
states the actual rule: the switch is named for the mode it turns on, the page it
reveals is "Agents", and the split is intentional. A test whose comment asserts an
invariant the code no longer holds is worse than no comment, so thank you for it.

@kyleseaman

Copy link
Copy Markdown
Contributor Author

FIXED — row deleted.

Correct, and verified before removing: developer.agents was derived only between
two edits inside this PR and never existed in any released build, so no saved link
or palette entry can carry it. A grep across website/src and src/kiro_crew
returns zero references after deletion.

The history comment above the map was wrong in the same way and is corrected: it
listed the never-shipped spelling as a step in the card's label history. The map
now documents that only ids a released build actually derived belong in it.
developer.crew-members and developer.crew-members-and-crew-mode did ship and
remain.

@kyleseaman

Copy link
Copy Markdown
Contributor Author

ACCEPTED — declared, and it follows the existing per-surface convention.

capture-agent-mode-card.mjs (and the roster/dialog/popover frames added for the
UX lane) is PR evidence tooling, not part of the rename. It is declared in the
body and matches the convention of the other per-surface capture scripts in
website/scripts/.

One design point on it, recorded here in case it is read as the stale-evidence
hazard a reviewer already raised: the script deletes each target frame BEFORE
driving the page, so every failure path leaves no frame. It does NOT delete a
frame written after the assertions passed, even if teardown later throws. That is
deliberate -- such a frame is correct evidence, and discarding it because
browser.close() hiccuped would destroy the artifact the script exists to
produce. An unremovable target fails loudly and names the file rather than
continuing over a stale frame.

@github-actions github-actions Bot added readiness: checking Automated validation is still running readiness: action required A blocking check or review needs attention merge conflict Branch has merge conflicts with its base — author must resolve before merge and removed readiness: action required A blocking check or review needs attention readiness: checking Automated validation is still running labels Sep 18, 2026
@NicholasRBowers NicholasRBowers added the needs-pr-triage PR scanner: awaiting automated triage label Sep 22, 2026
@dwu96 dwu96 added drive-to-green PR claimed by drive-to-green pipeline and removed needs-pr-triage PR scanner: awaiting automated triage labels Sep 22, 2026
@dwu96

dwu96 commented Sep 22, 2026

Copy link
Copy Markdown
Contributor

🤖 Kiro Crew [operator: dwu96#8c08bee0]: This PR has been inactive with failing CI and a merge conflict. I've assessed the blockers and they appear resolvable — I'll push fixes directly to this branch as a co-author.

Assessment: The only test failure is stale base drift, not a defect in this PR. test/test_ci_fleet_routing_expression_parity.py on this branch lists 9 entries in _EXPECTED_RESOLVER_CONSUMER_JOBS, while current main lists 19 (including ("ci.yml", "backend-test")) — and this PR touches no workflow or CI-parity file, so rebasing onto current main resolves the assertion by itself. Fix plan: rebase onto main, resolving the vocabulary-rename conflicts by taking main's structure and re-applying the Crew Members → Agents rename, then re-run the backend suites. Coverage Gate and PR Readiness are cascading from that one failing shard. No review lane is blocking — GPT 5.6 Review and Code Review both pass.

If you'd prefer I don't touch this PR, add the pr-no-autofix label.

@chenmingwei23

Copy link
Copy Markdown
Contributor

🤖 Kiro Crew [operator: chenmingwei23#de330d0c]: This PR has been inactive for 7+ days with a merge conflict blocking PR Readiness. I've assessed the blockers and they appear resolvable — I'll push fixes directly to this branch as a co-author.

Assessment: All AI review lanes (GPT, Opus, Design, First Principles, Security) pass; the only red is PR Readiness caused by a merge conflict against current main (39-file i18n rename "Crew Members → Agents"). Fix plan: rebase onto origin/main, take main's structure where the rename overlaps, re-run.

If you'd prefer I don't touch this PR, add the pr-no-autofix label.

@bolichen97

Copy link
Copy Markdown
Contributor

Closing: superseded by PR #12924 under the accepted Crewmates RFC

Thanks for the careful work here, especially the locale pronoun fixes. I'm closing this PR because, after it opened, the product settled on a different name for the same object, and PR #12924 now carries the same page copy using that name.

Overlap. Both PRs rewrite the same pages.membersPage.* and pages.kiroCrewAgentsPage.*_member values in en.json, en.manual.json and the locale catalogs, and the two sets of values can't both land:

Key this PR PR #12924
membersPage.title (en.json:7110) "Agents" "Crewmates"
membersPage.add_member (en.json:7106) "Add agent" "New crewmate"
membersPage.member_count_other "{{count}} agents" "{{count}} crewmates"

Why #12924 is the host

  • docs/request-for-change/rfc-crewmates-launch.md on main has status accepted and was merged in PR docs: add RFC for the Crewmates launch #12813 on 2026-09-22, four days after this PR opened. Its §3 makes "Crewmate" the product word, and its "Never say" column lists "agent (as the teammate)". It also says: "'Agent' is no longer a top-level concept in the product." feat(members): Crewmates page: empty hero, New crewmate dialog, MRU #12924 is in its implementation-prs. The "Agents" / "Agent Mode" instruction cited in this thread came before the RFC.
  • feat(members): Crewmates page: empty hero, New crewmate dialog, MRU #12924 deletes the ?new=1&from=members hand-off and its keys (create_member, pick_a_member, empty_roster). This PR renames those keys and re-pins them in CrewRoster.test.tsx and MembersPage.test.tsx.
  • This branch is 820 commits behind main and conflicts in 25 files. Several of its hunks rename keys that main has already removed: membersPage.crew_summary, edit_in_crew_manager, wake_none/wake_error and private_memory_legacy_confirm. A rebase would amount to a rewrite and would still ship copy the RFC rules out. The drive-to-green label invites exactly that kind of bot rebase.

Harvest. This is not a literal duplicate. Neither #12924 nor main touches the navigation surfaces yet. They should get a follow-up in crewmate vocabulary that reuses the mechanics here but not the values:

  • nav.crew_members (en.manual.json:3401) and the rail label at website/src/surfaces/builtins.tsx:79, plus CrewMemberMark.test.tsx
  • chatSidebar.open_crew_members / _desc (en.manual.json:4283), plus ChatSidebar.createMenu.test.tsx
  • featurePreviewsTab.crew_members / _desc (en.manual.json:6402-6403) and the intro-card strings
  • the settings card id move: SETTINGS_CREW_MEMBERS_PREVIEW_ID (website/src/hooks/useSettingHighlight.ts:19), the LEGACY_ID_EXACT compat rows (:56-57), and the regenerated settingsRegistry.gen.ts / settings-registry.generated.json

One change is vocabulary-neutral and can be ported as is: chatPage.this_session_cannot_be_opened_in_chat becomes "… — it belongs to {{surface}}." (en.json:6418), pinned at website/src/test/ChatPage.unresumableNotice.test.tsx:217-218.

website/scripts/capture-agent-mode-card.mjs isn't worth porting, since it only produces evidence for the "Agent Mode" frames.

If the RFC's vocabulary isn't meant to apply to this surface, reply here and reopen.

From a full open-PR audit against origin/main c66f95b (2026-09-25); findings were checked by independent reviewers reading the code. Reply here if this is wrong.

@bolichen97 bolichen97 closed this Sep 25, 2026
@github-actions github-actions Bot removed the readiness: action required A blocking check or review needs attention label Sep 25, 2026
@NicholasRBowers NicholasRBowers removed the drive-to-green PR claimed by drive-to-green pipeline label Sep 25, 2026
@bolichen97
bolichen97 deleted the chore/agents-vocabulary branch November 6, 2026 07:36
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

merge conflict Branch has merge conflicts with its base — author must resolve before merge

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants