Skip to content

feat(site/src): add an Owner scope toggle for shared workspaces - #29872

Merged
tracyjohnsonux merged 294 commits into
mainfrom
feat/filter-combobox-owner-shared-scope
Sep 29, 2026
Merged

tracyjohnsonux merged 294 commits into
mainfrom
feat/filter-combobox-owner-shared-scope

Conversation

@tracyjohnsonux

@tracyjohnsonux tracyjohnsonux commented Sep 24, 2026 •

Copy link
Copy Markdown
Collaborator

Summary

Replaces the separate User category on /workspaces (from #29586) with an "Include workspaces shared with {owner}" switch in the Owner flyout, so one Owner category covers both owner: and user:.

  • With no Owner chip the switch is on and disabled, so the first pick commits user:<name> (owned by or shared with that user). Once a chip is applied, the switch follows its key and later picks keep it. Turning the switch off commits owner:<name> (owned by that user). Flipping it rewrites the applied chip in place. A typed owner: or user: prefix sets the scope for that entry and reaches the query only when an option is picked. Without a typed prefix, an owner an applied chip holds, ignoring letter case, maps to that chip: selecting its option removes the chip (typed text highlights the first match, so Enter selects it), and typed Enter with no option highlighted keeps it. Otherwise a typed owner follows the switch when an option lists it; an unlisted owner goes under owner: unless user: was typed, because the backend rejects user: for users the requester cannot read (searchquery: user: filter rejects usernames the requester cannot read while owner: accepts them #29961).
  • While the switch is on, an include shared pill is joined to the Owner chip. The joined pill truncates on narrow widths so it stays inside the field, and hovering it shows the full switch label (for example, "Include workspaces shared with alice") on one line. Removing the pill (click, or Backspace on an empty input; screen readers hear "Hide workspaces shared with {owner}") turns the switch off. A second Backspace removes the Owner chip. Keyboard removal of the pill keeps focus in the search input.
  • Typing 3 or more characters from the start of shared with owner (for example sha) lists Owner and opens its flyout with the full list and the switch, until another row is highlighted. Picking an owner or flipping the switch there clears the typed text; text already applied as a search stays. Typing user also lists Owner.
  • A query with both an owner: and a user: token (such as a bookmarked user:me owner:alice) shows each chip under its own query key with no pill, so every edit keeps both. Selecting an owner that one of the chips holds, by click or by Enter on its highlighted row, removes that chip; typed Enter with no option highlighted keeps it. The switch is disabled in that state, because rewriting one chip to the other key would silently drop the other filter. A pick after a typed owner: or user: prefix replaces the chip under that key.
  • The current user's {username} (you) option is listed only when the typed text matches it (its label, me, or a users API match on name or email), like every other owner, so typing an unknown owner and pressing Enter commits that owner instead of me.
  • Removes the Shared attribute from "Workspace is…", an author decision after dogfooding: the switch is the menu path to shared workspaces. shared:true (any workspace with ACL entries) still works when typed.

No backend change: user: and owner: already exist in coderd/searchquery/search.go.

Stacked on #29876 (Clear all), the last of the PRs split out of #29426 to keep that PR focused on the mouse and touch redesign.

Decision notes

Generated with Coder Agents on behalf of @tracyjohnsonux.

…ouse-interactions

# Conflicts:
#	site/src/components/Filter/FilterCombobox/FilterCombobox.stories.tsx
#	site/src/components/Filter/FilterCombobox/FilterCombobox.tsx
#	site/src/components/Filter/FilterCombobox/filterQuery.ts
#	site/src/components/Filter/FilterCombobox/primitives.tsx
#	site/src/components/Filter/FilterCombobox/types.ts
#	site/src/components/Filter/FilterCombobox/useFilterCombobox.ts
#	site/src/pages/WorkspacesPage/filter/WorkspacesFilter.stories.tsx
#	site/src/pages/WorkspacesPage/filter/WorkspacesFilter.tsx
#	site/src/pages/WorkspacesPage/filter/categoryOptions.tsx
@tracyjohnsonux

Copy link
Copy Markdown
Collaborator Author

/coder-agents-review

@coder-agents-review

coder-agents-review Bot commented Sep 24, 2026 •

Copy link
Copy Markdown
Contributor

Chat: Review posted | View chat
Requested: 2026-09-26 11:09 UTC by @tracyjohnsonux

Review history
  • R1 (2026-09-24): 11 reviewers, 6 Nit, 4 P2, 13 P3, 4 P4, COMMENT. Review
  • R2 (2026-09-25), 6 Nit, 4 P2, 13 P3, 4 P4, COMMENT. Review
  • R3 (2026-09-25): 11 reviewers, 11 Nit, 5 P2, 19 P3, 13 P4, COMMENT. Review
  • R4 (2026-09-25), 11 Nit, 5 P2, 19 P3, 13 P4, COMMENT. Review
  • R5 (2026-09-25): 11 reviewers, 14 Nit, 7 P2, 25 P3, 14 P4, COMMENT. Review
  • R6 (2026-09-25), 14 Nit, 7 P2, 25 P3, 14 P4, COMMENT. Review
  • R7 (2026-09-25): 11 reviewers, 18 Nit, 7 P2, 28 P3, 25 P4, COMMENT. Review
  • R8 (2026-09-25): 11 reviewers, 20 Nit, 7 P2, 29 P3, 29 P4, COMMENT. Review
  • R9 (2026-09-26): 10 reviewers, 21 Nit, 8 P2, 30 P3, 32 P4, COMMENT. Review
  • R10 (2026-09-26): 11 reviewers, 23 Nit, 8 P2, 32 P3, 36 P4, COMMENT. Review
  • R11 (2026-09-26): 11 reviewers, 23 Nit, 8 P2, 32 P3, 39 P4, COMMENT. Review
  • R12 (2026-09-26): 10 reviewers, 23 Nit, 8 P2, 32 P3, 44 P4, COMMENT. Review
  • R13 (2026-09-26): 11 reviewers, 24 Nit, 8 P2, 32 P3, 49 P4, COMMENT. Review
  • R14 (2026-09-26): 11 reviewers, 26 Nit, 8 P2, 33 P3, 51 P4, COMMENT. Review
  • R15 (2026-09-26): 10 reviewers, 27 Nit, 8 P2, 33 P3, 55 P4, COMMENT. Review
  • R16 (2026-09-26): 10 reviewers, 30 Nit, 8 P2, 33 P3, 60 P4, COMMENT. Review
  • R17 (2026-09-26): 9 reviewers, 31 Nit, 8 P2, 33 P3, 60 P4, COMMENT. Review
  • R18 (2026-09-26): 5 reviewers, 31 Nit, 8 P2, 33 P3, 60 P4, APPROVE. Review
  • R19 (2026-09-26): 5 reviewers, 31 Nit, 8 P2, 33 P3, 60 P4, APPROVE. Review
  • R20 (2026-09-26): 8 reviewers, 32 Nit, 8 P2, 33 P3, 63 P4, COMMENT. Review
  • R21 (2026-09-26): 8 reviewers, 33 Nit, 8 P2, 33 P3, 63 P4, COMMENT. Review
  • R22 (2026-09-26): 6 reviewers, 33 Nit, 8 P2, 33 P3, 63 P4, APPROVE. Review

deep-review v0.13.0 | Round 22 | 219e83e..1142d5e

Last posted: Round 22, 137 findings (8 P2, 33 P3, 63 P4, 33 Nit), APPROVE. Review

Finding inventory

Finding inventory: PR #29872

Findings

# Sev Status Location Summary Round Reviewer Posted
CRF-1 P2 Author fixed (9f90a12) useFilterCombobox.ts:659 Removing the scope pill deletes typed search text R1 Netero Yes
CRF-2 P2 Author fixed (9f90a12, 196840d) useFilterCombobox.ts:905 Typed owner: prefix commits a user: chip R1 Netero Yes
CRF-3 P3 Author fixed (9f90a12) WorkspacesFilter.tsx:62 Typing user: no longer opens user suggestions R1 Netero Yes
CRF-4 P3 Author fixed (9f90a12) types.ts:14 token doc comment still cites shared:true R1 Netero Yes
CRF-5 P4 Author fixed (9f90a12) useFilterCombobox.ts:613 withOptionToken other-key drop has no test R1 Netero Yes
CRF-6 Nit Author fixed (9f90a12, ab755ae) WorkspacesFilter.stories.tsx:66 Edited play functions assert toBeVisible() R1 Netero Yes
CRF-7 Note Note (body) categoryOptions.tsx:167 Existing shared:true URLs render as free text R1 Netero No
CRF-8 P2 Author accepted R5 (human author confirmed on the thread that losing the any-sharing menu path is accepted; description updated) categoryOptions.tsx:-161 Removing Shared attribute drops the any-sharing filter; switch does not replace it R1 Pariston P2, Luffy P2 Yes
CRF-9 P2 Author fixed (9f90a12) useFilterCombobox.ts:611 Mixed owner:/user: query shows two Owner chips; pill, Backspace, switch act on wrong chip R1 Meruem P2, Melody P3, Mafuuu P3 Yes
CRF-10 P3 Author fixed (886074f) useFilterCombobox.ts:272 No-chip switch state depends on removal history; untested R1 Nami P3, Bisky P3, Pariston P4, Meruem P4, Leorio P4 Yes
CRF-11 P3 Author fixed (9f90a12) useFilterCombobox.ts:368 Scope typeahead hijacks sha/with searches; Enter clears them R1 Luffy P3, Meruem P3 Yes
CRF-12 P3 Acknowledged R4 (author decision: default stays on; the pill shows the widened scope); Luffy, who raised it, did not re-raise R5 useFilterCombobox.ts:288 Switch on by default: owner pick includes workspaces owned by others R1 Luffy Yes
CRF-13 P3 Author fixed (886074f) WorkspacesFilter.tsx:69 Pill text reads as a narrowing condition R1 Leorio Yes
CRF-14 P3 Author fixed (886074f) WorkspacesFilter.tsx:67 No-chip switch label promises a list change that does not happen R1 Leorio Yes
CRF-15 P3 Author fixed (886074f) FilterCombobox.tsx:408 Truncated pill text unreachable on touch and keyboard R1 Nami Yes
CRF-16 P3 Author fixed (9f90a12) types.ts:57 chipKey-in-chipKeys invariant enforced only by doc; picks silently dropped R1 Meruem P3, Mafuuu P4 Yes
CRF-17 P3 Author fixed (9f90a12) filterQuery.ts:78 New chipDisplay branch has no unit test R1 Bisky Yes
CRF-18 P4 Author fixed (9f90a12, 7ed2266) useFilterCombobox.ts:512 Suggestion and typed-Enter commit key untested R1 Bisky Yes
CRF-19 P4 Author fixed (9f90a12) FilterCombobox.tsx:816 splitIntoTwoLines duplicates text-balance R1 Nami P4, Luffy P4 Yes
CRF-20 P4 Author fixed (9f90a12) types.ts:64 pillLabel undocumented lowercasing and typeahead role R1 Leorio Yes
CRF-21 P3 Author fixed (9f90a12) types.ts:63 scopeToggle.chipKey name collides with other chipKey meanings R1 Gon Yes
CRF-22 P3 Author fixed (9f90a12) FilterCombobox.tsx:145 Comments call the pill label the scope toggle label R1 Gon P3, Leorio P4 Yes
CRF-23 P3 Author fixed (9f90a12) useFilterCombobox.ts:634 typedFilterText boolean naming; comment omits toggleScope consumer R1 Gon Yes
CRF-24 Nit Author fixed (9f90a12) useFilterCombobox.ts:369 Unnamed min length 3; scope-match comment omits rules R1 Gon P3 + P2 Yes
CRF-25 Nit Author fixed (9f90a12) useFilterCombobox.ts:1014 scopeWidened comment describes optionChipKey R1 Gon P2, Leorio Nit Yes
CRF-26 Nit Author fixed (9f90a12) useFilterCombobox.ts:290 optionChipKey comment restates the code R1 Gon P2 Yes
CRF-27 Nit Author fixed (9f90a12) WorkspacesFilter.stories.tsx:63 Story comment says user:me, assertion says owner:me R1 Leorio Yes
CRF-28 Nit Author fixed (9f90a12) categoryOptions.tsx:99 Stale comments name separate User and Owner categories R1 Melody Nit, Luffy Nit Yes
CRF-29 Note Note (body) types.ts:60 Generic scopeToggle option has one consumer R1 Pariston No
CRF-30 P1 Dropped by orchestrator (disproved: composeFilterQuery dedupes same-key tokens; scratch test emitted user:alice) useFilterCombobox.ts:611 Picking an owner on user:me emits user:me user:alice R1 Mafuuu No
CRF-31 P2 Dropped by orchestrator (same-key case collapses downstream; mixed-key gap is CRF-9, whose fix makes the comment true) useFilterCombobox.ts:611 withOptionToken comment says one chip R1 Gon No
CRF-32 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.tsx:859 Keyboard users can reach the flyout switch R1 Nami No
CRF-33 Note Dropped by orchestrator (praise, no change requested) WorkspacesFilter.tsx:59 Merging User and Owner is an improvement R1 Luffy No
CRF-34 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:290 Scope toggle pairings agree for one chip R1 Melody No
CRF-35 P2 Dropped by orchestrator (illustrative example in a type doc; no rule forbids it) types.ts:52 scopeToggle doc carries a Workspaces example R1 Gon No
CRF-36 P2 Dropped by orchestrator (doc already states the reason Gon asks to keep) FilterCombobox.tsx:813 splitIntoTwoLines doc restates the loop R1 Gon No
CRF-37 P2 Dropped by orchestrator (pre-existing comment kept accurate by the PR, per AGENTS.md preserve-comments rule) FilterCombobox.tsx:886 OptionsPanel comment lists its slots R1 Gon No
CRF-38 Nit Dropped by orchestrator (checked is the Switch API; on/off names the UI control) FilterCombobox.tsx:810 Three vocabularies for one switch state R1 Gon No
CRF-39 Note Dropped by orchestrator (summary of Gon's other findings) useFilterCombobox.ts:290 Comment quality pattern R1 Gon No
CRF-40 OOS Out of scope (body) useFilterCombobox.ts:957 Enter on a text-matched category row clears the free-text search R1 Nami, Meruem No
CRF-41 OOS Out of scope (body) filterQuery.ts:29 categoryPreview has no production caller and uses category.key R1 Melody No
CRF-42 OOS Out of scope (body) coderd/searchquery/search.go:283 Workspace query language has no OR R1 Pariston No
CRF-43 OOS Out of scope (body) WorkspacesFilter.tsx:53 Comment repeats the TODO above it (predates the PR) R1 Gon No
CRF-44 OOS Dropped by orchestrator (disproved: dedupeInOrder collapses same-key tokens) useFilterCombobox.ts:685 Picking two templates emits duplicate template: keys R1 Mafuuu No
CRF-45 P3 Author fixed (7ed2266) useFilterCombobox.ts:681 withOptionToken duplicates the dedupeChips Owner collapse; its test passes without it R3 Netero Yes
CRF-46 P3 Author fixed (7ed2266) useFilterCombobox.ts:959 Typed-prefix branch reimplements toggleScope rewrite and narrowedScopes update R3 Netero Yes
CRF-47 P4 Author fixed (7ed2266) WorkspacesFilter.tsx:61 chipKeys: ["owner", "user"] repeats widenedKey R3 Netero Yes
CRF-48 P4 Author fixed (7ed2266) filterQuery.ts:66 scopeGroupKey and chipDisplay re-derive the key set instead of categoryChipKeys R3 Netero Yes
CRF-49 P4 Author fixed (7ed2266) useFilterCombobox.ts:1107 Backspace passes clearCategorySearch = true, which cannot apply R3 Netero Yes
CRF-50 P4 Author fixed (7ed2266) useFilterCombobox.ts:773 Search-text clear in toggleCategoryOption removal branch untested R3 Netero Yes
CRF-51 Nit Author fixed (7ed2266) filterQuery.ts:234 typedKey re-parsed with dead ?. and ?? "" R3 Netero Yes
CRF-52 P2 Author fixed (7ed2266) filterQuery.ts:66 owner:+user: collapse drops a filter the backend still applies; next edit removes it R3 Mafuuu P2, Knov P2, Pariston P3, Melody P3 Yes
CRF-53 P3 Author fixed (7ed2266) useFilterCombobox.ts:959 Typed owner:/user: prefix rewrites the chip and emits before a pick; Escape does not undo R3 Melody Yes
CRF-54 P4 Author fixed (7ed2266) FilterCombobox.tsx:1173 Drill-in error branch hides the scope switch; flyout shows it R3 Nami Yes
CRF-55 P4 Author fixed (7ed2266) types.ts:56 scopeToggle accepted on inline categories but inline rows ignore it R3 Melody Yes
CRF-56 P4 Author fixed (7ed2266) FilterCombobox.tsx:393 Pill scan and scopeValue handle two Owner chips, which cannot occur R3 Knov P4, Bisky P4 Yes
CRF-57 P4 Author fixed (7ed2266) useFilterCombobox.ts:31 SCOPE_MATCH_MIN_QUERY_LENGTH not pinned by a test R3 Bisky Yes
CRF-58 P3 Author fixed (7ed2266) filterQuery.ts:52 Dedupe docs describe per-key collapse; scope grouping and categories undocumented R3 Gon P3, Leorio P4 Yes
CRF-59 P3 Author fixed (7ed2266) filterQuery.ts:23 chipDisplay doc rule contradicted by the widenedKey branch R3 Gon P3, Leorio P4 Yes
CRF-60 Nit Author fixed (7ed2266) FilterCombobox.tsx:166 Scope-match rule copied into four comments; this one omits its conditions R3 Gon P2, Leorio Nit Yes
CRF-61 Nit Author fixed (7ed2266) filterQuery.ts:181 CategoryMatchSource.aliases doc repeats FilterCategory.aliases R3 Gon P2 Yes
CRF-62 Nit Author fixed (7ed2266) useFilterCombobox.ts:426 scopeMatchOnlyCategory boolean named like a category R3 Gon Yes
CRF-63 Nit Author fixed (7ed2266) FilterCombobox.tsx:251 toggleScope(categoryKey, true) unlabeled boolean R3 Gon Yes
CRF-64 P3 Author fixed (PR description updated) useFilterCombobox.ts:413 PR description says word-start match and two-line tooltip; both changed R3 Leorio P3, Gon Note Yes
CRF-65 P4 Author fixed (7ed2266) types.ts:24 aliases doc example "user for owner" contradicts widenedKey design R3 Gon (listed as OOS; moved in scope) Yes
CRF-66 P3 Dropped by orchestrator (in-scope field follows pre-existing panel chipKey naming that R1 accepted as the current-key meaning) filterQuery.ts:184 chipKey fields collide with chipKeys R3 Gon No
CRF-67 OOS Out of scope (body) FilterCombobox.tsx:1031 Pre-existing panel chipKey props sit next to chipKeys R3 Gon No
CRF-68 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.tsx:1009 Switch covers loading, empty, flyout-error states R3 Nami No
CRF-69 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.tsx:394 Pill-render and Backspace rules agree R3 Melody No
CRF-70 Note Dropped by orchestrator (praise, no change requested) WorkspacesFilter.tsx:65 Switch label and decision notes work R3 Leorio No
CRF-71 P2 Author fixed (0ea86e5) useFilterCombobox.ts:999 Editing the input after the kept second Owner token drops that token's filter R5 Netero Yes
CRF-72 P3 Deferred (#29955) docs/user-guides/shared-workspaces.md:85 User guide still says to filter for "Shared" R5 Netero Yes
CRF-73 Note Note (body) filterQuery.ts:110 Repeated scope key keeps the first token as chip, unlike other keys R5 Netero No
CRF-74 P3 Author fixed (0ea86e5) primitives.tsx:449 Base merge restores keyboard focus move on every chip removal; untested R5 Law Yes
CRF-75 P2 Author fixed (0ea86e5) useFilterCombobox.ts:765 Switch or pill with a kept second Owner token emits a repeated key the backend rejects R5 Mafuuu P2, Knov P2, Pariston P2 Yes
CRF-76 P3 Author fixed (0ea86e5) FilterCombobox.tsx:401 Pill follows the typed prefix, not the applied chip, so it contradicts the query R5 Melody Yes
CRF-77 P3 Author fixed (0ea86e5) filterQuery.ts:172 Optional categories = [] lets callers drop the kept token; three call sites untested R5 Bisky Yes
CRF-78 P4 Author fixed (0ea86e5) useFilterCombobox.ts:746 hasChip guard in toggleScope cannot run and repeats the switch disabled rule R5 Bisky Yes
CRF-79 P3 Author fixed (0ea86e5) types.ts:55 scopeToggle doc and PR description say each pick turns the switch on R5 Mafuuu P3, Gon P3, Leorio P3, Luffy P4 Yes
CRF-80 P3 Author fixed (0ea86e5) useFilterCombobox.ts:1298 scopeMatchKey and story comments say the match is on the pill label R5 Gon P3, Leorio P4 Yes
CRF-81 Nit Author fixed (0ea86e5) FilterCombobox.tsx:399 Local scopeToggle means the pill's toggle; comment above it adds nothing R5 Gon P2 Yes
CRF-82 Nit Author fixed (0ea86e5) filterQuery.ts:72 scopeCategoryKey comment restates its body R5 Gon P2 Yes
CRF-83 P2 Dropped by orchestrator (comment is true: the options serve both owner: and user: tokens, which is why they carry no key) categoryOptions.tsx:117 getUserFilterOptions comment says "shared by" with one caller R5 Gon No
CRF-84 Nit Author fixed (0ea86e5) types.ts:78 Exported categoryChipKeys has no doc comment R5 Leorio Yes
CRF-85 Note Dropped by orchestrator (observation, no change requested) WorkspacesFilter.tsx:55 One Owner category runs one options query per keystroke R5 Nami No
CRF-86 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.tsx:1026 Switch renders in every panel state; state depends on chip and typed prefix R5 Nami No
CRF-87 Note Dropped by orchestrator (praise, no change requested) WorkspacesFilter.tsx:55 One Owner category with a pill works better than two categories R5 Luffy No
CRF-88 Note Dropped by orchestrator (cleared concern, no change requested) filterQuery.ts:115 owner/user key pair agrees across consumers R5 Melody No
CRF-89 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.test.tsx:1519 New tests fail when their behavior is removed R5 Bisky No
CRF-90 P4 Author fixed (64dbfce) FilterCombobox.stories.tsx:683 ScopeToggleWithSecondOwner play checks neither the disabled switch nor the hidden pill button R7 Netero Yes
CRF-91 P4 Author fixed (64dbfce) useFilterCombobox.ts:1218 Backspace pill branch re-finds a category scopePillCategoryKey already named R7 Netero Yes
CRF-92 P4 Author fixed (64dbfce) useFilterCombobox.ts:1373 Six copies of the category-by-key lookup; scopeFor runs it four times R7 Netero P4, Zoro P4 Yes
CRF-93 P3 Author fixed (64dbfce) useFilterCombobox.ts:821 Typed-prefix pick on a two-chip Owner query replaces both chips with one R7 Nami Yes
CRF-94 P3 Author fixed (64dbfce) FilterCombobox.tsx:442 Mixed query shows two owner: chips and a pill with no remove button; nothing says why the switch is disabled R7 Luffy P3, Nami P4, Leorio P4 Yes
CRF-95 P3 Author fixed (64dbfce) useFilterCombobox.ts:763 Scope flyout switch or pick deletes an applied free-text search that matches the scope phrase R7 Melody Yes
CRF-96 P4 Author fixed (64dbfce) useFilterCombobox.ts:1093 Typed alias or label of a scope category maps to widenedKey R7 Melody P4, Zoro P4 Yes
CRF-97 P4 Author fixed (64dbfce) useFilterCombobox.ts:1080 Typing user without a colon lists no category; base listed User R7 Mafuuu Yes
CRF-98 P4 Author fixed (64dbfce) types.ts:59 scopeToggle doc says the first pick turns the switch on (it is already on) and omits the typed prefix R7 Leorio P4, Mafuuu P4 Yes
CRF-99 P4 Author fixed (64dbfce) FilterCombobox.test.tsx:1899 Test asserts a repeated-key query dedupe cannot emit; comment and description give the wrong reason for the guard R7 Bisky Yes
CRF-100 P4 Author fixed (64dbfce) FilterCombobox.test.tsx:1965 Flyout pin after flipping the switch from a scope match is untested R7 Bisky Yes
CRF-101 P4 Author fixed (64dbfce) FilterCombobox.stories.tsx:711 ScopePillTooltip does not check the tooltip text R7 Bisky Yes
CRF-102 P4 Author fixed (64dbfce) useFilterCombobox.ts:327 typedScope stores a repeated category key and a key only compared to widenedKey R7 Zoro Yes
CRF-103 P4 Author fixed (64dbfce) useFilterCombobox.ts:884 toggleCategoryOption reimplements toggledChips R7 Zoro Yes
CRF-104 Nit Author fixed (64dbfce) useFilterCombobox.ts:1392 isPlainScopeChip is true for chips that are not the scope chip R7 Gon P3 Yes
CRF-105 Nit Author fixed (64dbfce) useFilterCombobox.ts:1377 scopeValue and optionChipKey return-field comments restate their names R7 Gon P2 Yes
CRF-106 Nit Author fixed (64dbfce) useFilterCombobox.ts:792 rewriteScopeKey takes an unlabeled boolean R7 Gon Yes
CRF-107 Nit Author fixed (64dbfce) FilterCombobox.stories.tsx:705 Story comment calls the switch label "full toggle message" R7 Gon Yes
CRF-108 Note Dropped by orchestrator (praise, no change requested) filterQuery.ts:57 filterQuery.ts is back to one chip per key R7 Luffy No
CRF-109 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:821 Four option-commit paths agree R7 Melody No
CRF-110 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:343 Repeated-key guard covers every scope rewrite path R7 Mafuuu No
CRF-111 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:344 categories removal and shared predicate fix the round 5 test findings R7 Bisky No
CRF-112 P4 Author fixed (5dab005) FilterCombobox.test.tsx:1851 No test pins that the first chip of user:me owner:carol shows user:me R8 Netero Yes
CRF-113 P3 Author fixed (5dab005) useFilterCombobox.ts:386 With a chip under each Owner key, the owner list checks only the first chip's key; clicking the second chip's owner replaces the first R8 Mafuuu P3, Melody P4, Nami P4 Yes
CRF-114 P4 Author fixed (5dab005) types.ts:61 scopeToggle doc and hook comment omit that a typed alias commits under the category key R8 Leorio P4, Gon P4 Yes
CRF-115 P4 Author fixed (5dab005) FilterCombobox.test.tsx:1958 Typed-alias test starts from a narrowed chip, so it cannot tell narrowing from deferring R8 Bisky P4, Leorio P4 Yes
CRF-116 Nit Author fixed (5dab005) FilterCombobox.test.tsx:1901 Test name "replaces the Owner chip under the typed key" is false for its owner:bob case R8 Leorio Yes
CRF-117 P4 Author fixed (5dab005) FilterCombobox.tsx:251 Comment says a click enters the category like Enter, but Enter on a scope-only row applies a search; click path untested R8 Razor Yes
CRF-118 Nit Author fixed (5dab005) useFilterCombobox.ts:1364 scopeState comment restates its name and guard R8 Gon P2 Yes
CRF-119 P2 Dropped by orchestrator (the comment gives the state's role and the typed-prefix override; its alias gap is part of CRF-114) useFilterCombobox.ts:344 typedScopeWidened comment restates isScopeWidened R8 Gon No
CRF-120 P2 Dropped by orchestrator (disproved: the comment's second sentence states the consequence, "the query would silently lose a filter") useFilterCombobox.ts:359 isScopeToggleDisabled comment restates the count and does not say what a collision does R8 Gon No
CRF-121 P2 Dropped by orchestrator (the comment already says when the fallback runs, which is the part Gon asks to keep) useFilterCombobox.ts:820 withCategoryOption comment restates its expression R8 Gon No
CRF-122 OOS Out of scope (body) WorkspacesFilter.tsx:45 canListUsers checks viewDeploymentConfig, not a list-users capability R8 Razor No
CRF-123 OOS Out of scope (body) types.ts:24 aliases doc omits that aliases also match typed text with no colon R8 Gon No
CRF-124 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:369 Scope pairings changed in 64dbfce agree R8 Melody No
CRF-125 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:238 Alias memo feeds all three typed-text consumers R8 Mafuuu No
CRF-126 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.test.tsx:2074 Round 7 test fixes fail when their behavior is removed R8 Bisky No
CRF-127 P3 Author fixed (92c95ec) useFilterCombobox.ts:694 Two-key state: value suggestions and typed Enter commit under the first chip's key and replace that chip R9 Netero P3, Melody P3, Razor P4 Yes
CRF-128 P4 Author fixed (92c95ec) useFilterCombobox.ts:400 categoryOptionToken ignores a typed prefix; typed pick on a two-key query removes the other key's chip R9 Netero Yes
CRF-129 P2 Author fixed (92c95ec) useFilterCombobox.ts:1251 Typed Enter in Owner commits user:<typed>, which the backend rejects for members who cannot read that user R9 Chopper Yes
CRF-130 P4 Author fixed (92c95ec) FilterCombobox.tsx:582 Two-key selection fix is pinned only for the drill-in click; flyout click and Enter highlight check are untested R9 Bisky P4, Mafuuu P4 Yes
CRF-131 P4 Author fixed (92c95ec) types.ts:64 scopeToggle doc says a pill shows while the switch is on; with a chip under each key none does R9 Leorio Yes
CRF-132 Nit Author fixed (92c95ec) useFilterCombobox.ts:395 One function named categoryOptionToken inside the hook and optionTokenFor in its return value R9 Gon Yes
CRF-133 P2 Dropped by orchestrator (round 8 asked for this sentence at this location in CRF-117; Tab completes like Enter throughout the hook) FilterCombobox.tsx:252 Scope-row Enter sentence describes handleKeyDown and omits Tab R9 Gon No
CRF-134 P2 Dropped by orchestrator (the comment already states the consequence: the row shows selected and a click removes the chip) useFilterCombobox.ts:393 categoryOptionToken comment restates its expression R9 Gon No
CRF-135 OOS Out of scope (body) coderd/searchquery/search.go:728 user: returns 400 for usernames the requester cannot read; owner: accepts any string R9 Chopper No
CRF-136 OOS Out of scope (body) FilterCombobox.tsx:670 Applied-option check mark is aria-hidden and rows carry no selected state for screen readers R9 Nami No
CRF-137 Note Dropped by orchestrator (cleared concern, no change requested) WorkspacesFilter.test.tsx:52 Permission-set test checks the default user:me outcome R9 Chopper No
CRF-138 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.tsx:1052 optionTokenFor prop doc is accurate R9 Leorio No
CRF-139 Note Dropped by orchestrator (evidence folded into CRF-130) useFilterCombobox.ts:1256 Reverting the Enter highlight check fails no test R9 Bisky No
CRF-140 P3 Author fixed (3302857) useFilterCombobox.ts:1293 Typed Enter commits under owner: for listed owners too, so Enter narrows what a click on the same row widens R10 Netero Yes
CRF-141 P4 Author fixed (3302857) useFilterCombobox.ts:1294 No test covers appliedScopeChipFor in the typed-Enter candidate R10 Netero Yes
CRF-142 P4 Author fixed (3302857) types.ts:55 scopeToggle doc omits that typed Enter commits under the category key unless widenedKey was typed R10 Netero Yes
CRF-143 P3 Author fixed (3302857) useFilterCombobox.ts:1292 Typed Enter on an owner an applied chip holds keeps the chip; PR description and appliedScopeChipFor comment describe removal R10 Leorio P3, Razor P4, Gon P4 Yes
CRF-144 P4 Author fixed (3302857) useFilterCombobox.ts:1288 Typed-Enter comment states a Workspaces backend workaround in Owner terms inside the generic hook, without linking #29961 R10 Mafuuu P4, Zoro P4, Leorio P4, Razor Nit Yes
CRF-145 P4 Author fixed (3302857) useFilterCombobox.ts:1293 Typed-Enter candidate has two branches that both run appliedScopeChipFor and differ only in the key R10 Zoro Yes
CRF-146 Nit Author fixed (3302857) filterQuery.ts:178 CategoryMatchSource.optionToken shares its name with the module function optionToken and is a third name for optionTokenFor R10 Gon, Leorio Yes
CRF-147 Nit Author fixed (3302857) FilterCombobox.test.tsx:40 Fixtures scopedOwnersCategory and scopedOwnerCategory differ by one letter that does not name their difference R10 Gon Yes
CRF-148 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:409 Every Owner option-token producer and consumer goes through optionTokenFor R10 Melody No
CRF-149 Note Dropped by orchestrator (cleared concern, no change requested) filterQuery.ts:178 optionToken callback in place of chipKey is justified R10 Zoro No
CRF-150 P4 Author fixed (cee3c66) useFilterCombobox.ts:1357 listed matches the typed owner with exact case, so typed Enter on a listed owner in another case still commits under owner: R11 Netero Yes
CRF-151 P4 Author fixed (cee3c66) FilterCombobox.test.tsx:2257 "keeps the applied chip holding a typed unlisted owner" never asserts that user:zed is kept; Enter removing it passes R11 Leorio P4, Bisky P4 Yes
CRF-152 P4 Author fixed (cee3c66) types.ts:63 Typed-Enter rule in scopeToggle doc, hook comment, and PR description omits that an applied chip holding the value is kept; description says highlighting alone removes a chip R11 Razor P4, Gon P4, Mafuuu P4, Leorio P4 Yes
CRF-153 P2 Dropped by orchestrator (round 10 asked for the rule in this comment in CRF-144; the drift it causes is CRF-152, whose fix offers Gon's shorter comment) useFilterCombobox.ts:1353 Typed-Enter comment restates the ternary and repeats the types.ts rule R11 Gon No
CRF-154 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:1352 Typed-Enter comment and scopeToggle doc name rule, reason, and #29961 R11 Leorio No
CRF-155 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:1356 listed branch is pinned by a test; categories without a scope toggle unchanged R11 Razor No
CRF-156 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:404 CRF-143 and CRF-144 fixes match the code R11 Razor No
CRF-157 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:1359 candidate is one expression and listed is tested R11 Zoro No
CRF-158 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:1356 Listed owners are users the requester can read, so user: is safe for them R11 Knov No
CRF-159 P4 Author fixed (e3352f4) useFilterCombobox.ts:418 appliedScopeChipFor compares chip and option values with exact case; with both Owner keys applied, a case-different chip is not matched and the other chip is replaced R12 Netero Yes
CRF-160 P4 Author fixed (e3352f4) useFilterCombobox.ts:1364 No test covers the applied-chip lookup inside optionTokenFor on the listed typed-Enter branch; removing it drops owner:me with all tests passing R12 Bisky Yes
CRF-161 P4 Author fixed (e3352f4) types.ts:65 Doc and PR description say an applied chip holding an unlisted typed value keeps its key, but with a single user: chip Enter commits owner: R12 Razor P4, Mafuuu P4 Yes
CRF-162 P4 Author fixed (e3352f4) useFilterCombobox.ts:1346 Comment says typed Enter commits the typed value; since cee3c66 a case-insensitive listed match commits the option's token, in every submenu category R12 Gon P4, Leorio P4, Mafuuu P4 Yes
CRF-163 P4 Author fixed (e3352f4) useFilterCombobox.ts:364 typedScopeWidened is useState outside the reducer and stays consistent with activeCategoryKey only through hand-paired setter calls R12 Meruem Yes
CRF-164 Note Dropped by orchestrator (cleared concern; its CRF-152 part is contradicted by the scratch tests in CRF-161) FilterCombobox.test.tsx:2328 CRF-151 and CRF-152 fixed R12 Leorio No
CRF-165 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.test.tsx:2292 CRF-150 and CRF-151 tests fail when their behavior is removed R12 Bisky No
CRF-166 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:1357 Case-insensitive listed match agrees with every Workspaces backend lookup R12 Razor No
CRF-167 Note Dropped by orchestrator (evidence folded into CRF-162) useFilterCombobox.ts:1357 Case-insensitive listed match applies to every category, not only Owner R12 Meruem No
CRF-168 P4 Author fixed (458fbb9) FilterCombobox.test.tsx:2456 The zed row of "commits typed %s holding the only Owner chip user:zed" passes when Enter commits nothing, because entering the category already emits user:zed R13 Netero Yes
CRF-169 P4 Author fixed (458fbb9) useFilterCombobox.ts:447 appliedScopeChipFor matches ignoring case only with two chips; with one user:Alice chip the alice row is unchecked and a click rewrites the chip, while typed Enter keeps it R13 Meruem P4, Melody P4 Yes
CRF-170 P4 Author fixed (458fbb9) useFilterCombobox.ts:1417 Listed typed-Enter branch calls optionTokenFor, whose applied-chip lookup cannot return a chip after scopeChipHolding ran; replacing it passes all 205 tests R13 Bisky Yes
CRF-171 P4 Author fixed (458fbb9) useFilterCombobox.ts:388 Scope-decision comment lost its useState in e3352f4 and now sits above chipKeyOf, not isScopeWidened R13 Gon P2, Razor Nit, Mafuuu Nit, Leorio Nit Yes
CRF-172 P4 Author fixed (458fbb9) types.ts:64 "in any case" reads as "always", not "ignoring letter case"; the same rule has three phrasings across doc, hook comment, description, and a test name R13 Gon P4, Razor Nit, Leorio Nit Yes
CRF-173 Nit Author fixed (458fbb9) FilterCombobox.test.tsx:2435 Test name says the typed text holds the chip; the chip holds the typed value R13 Gon, Leorio Yes
CRF-174 P2 Dropped by orchestrator (the comment states the function's contract; the case fold is pinned by two tests, Netero M1) useFilterCombobox.ts:430 scopeChipHolding comment restates its body and omits why values match ignoring case R13 Gon No
CRF-175 P4 Dropped by orchestrator (round 12 asked for these resets in CRF-163; they keep the field null outside a category) useFilterCombobox.ts:157 Seven typedScopeWidened: null resets cannot change behavior R13 Razor No
CRF-176 OOS Out of scope (body) FilterCombobox.tsx:1221 Row selection uses exact-case selectedTokens.includes(token) in every submenu category (unchanged from base) R13 Mafuuu, Melody No
CRF-177 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.test.tsx:2387 e3352f4 tests fail when their behavior is removed R13 Bisky No
CRF-178 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:1414 CRF-161 to CRF-163 fixes hold R13 Mafuuu No
CRF-179 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:62 CRF-163 fix verified R13 Meruem No
CRF-180 Note Dropped by orchestrator (cleared concern, no change requested) WorkspacesFilter.test.tsx:52 Permission-set test covers both Owner option loaders R13 Nami No
CRF-181 P3 Author fixed (8a1ce33, 9a045a4) FilterCombobox.tsx:185 Typing sha shows the Owner flyout only if Owner was the last highlighted category row and no other category's flyout is open R14 Netero P3, Nami P3 Yes
CRF-182 Nit Author fixed (8a1ce33, 9a045a4) useFilterCombobox.ts:1390 Comment line not rewrapped after "ignoring letter case" edit (84 characters wide) R14 Netero Yes
CRF-183 P4 Author fixed (8a1ce33, 9a045a4) FilterCombobox.tsx:179 shownFlyoutKey comment and highlightedCategoryKey name say current highlight; the value is the last highlighted category row, kept when the highlight clears R14 Gon P4, Leorio P4 Yes
CRF-184 P4 Author fixed (8a1ce33, 9a045a4) useFilterCombobox.ts:404 isScopeWidened comment lost the typed-prefix sentence in the CRF-171 move, so it says the chip decides while the first branch returns the typed prefix R14 Razor P4, Mafuuu P4, Leorio P4 Yes
CRF-185 Nit Author fixed (8a1ce33, 9a045a4) FilterCombobox.test.tsx:2536 Test name "removes the chip of %s holding ..." renders the whole two-chip query as one chip R14 Gon, Leorio Yes
CRF-186 P2 Dropped by orchestrator (contradicted by CRF-184: the comment needs the decision order, including the chip rule; same pattern as CRF-174) useFilterCombobox.ts:404 isScopeWidened comment's first sentence restates the body R14 Gon No
CRF-187 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:445 Every Owner token producer and consumer agrees for a single chip in different letter case R14 Melody No
CRF-188 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:1410 CRF-168 to CRF-170 fixes complete and pinned R14 Bisky No
CRF-189 Note Dropped by orchestrator (evidence folded into CRF-181) FilterCombobox.test.tsx:2825 All four scope-flyout tests render only Owner, so Owner is highlighted before typing R14 Bisky No
CRF-190 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:449 CRF-168 to CRF-170 fixes hold R14 Mafuuu No
CRF-191 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:449 CRF-169 fix holds for every optionTokenFor consumer R14 Razor No
CRF-192 P4 Author fixed (710a26d) FilterCombobox.tsx:287 useFlyoutOptions narrowed-text condition and its comment clause at :281 restate the shownFlyoutKey rule; the condition cannot change the result since 9a045a4 R15 Netero P4, Gon Nit Yes
CRF-193 P4 Author fixed (710a26d) FilterCombobox.tsx:183 Comment says null means "a dismissed menu", but close never sets it; the other writer, onMouseLeave (:609), hides the typed sha flyout, which the PR description says stays until another row is highlighted R15 Gon P4, Meruem P4, Razor P4 Yes
CRF-194 P4 Author fixed (710a26d) FilterCombobox.test.tsx:2952 "hovering Running" row's user.type clicks the input and fires the menu's onMouseLeave, so it repeats the "leaving the menu" row and typing with the pointer on another row is untested R15 Bisky Yes
CRF-195 P4 Author fixed (710a26d) FilterCombobox.tsx:195 No test covers another highlighted row hiding the scope-match flyout; dropping the highlight condition passes 222 of 222 tests R15 Bisky Yes
CRF-196 Nit Author fixed (710a26d) FilterCombobox.test.tsx:2914 Test name "after highlighting another row" does not separate it from the it.each rows that also highlight another row; it presses End R15 Gon Yes
CRF-197 P2 Dropped by orchestrator R16 (reopened R16; the false condition 710a26d introduced is CRF-204; the restating and duplication parts stay dropped: the comment states the rule across narrowing, highlight, and pointer type; CRF-174, CRF-186 precedent) FilterCombobox.tsx:188 shownFlyoutKey comment restates the ternary and repeats the "flyout returns when text is deleted" rule from :169-170 and :265-266 R15 Gon No
CRF-198 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:441 CRF-184 fixed R15 Razor No
CRF-199 Note Dropped by orchestrator (cleared concern, no change requested; its "without leaving" coverage claim is contradicted by CRF-194) FilterCombobox.tsx:197 CRF-181's two causes fixed, each with a test R15 Razor No
CRF-200 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.test.tsx:2965 CRF-181 tests other than the "hovering Running" row fail when their fixes are reverted R15 Bisky No
CRF-201 P4 Author fixed (43743c8) FilterCombobox.tsx:291 Merge d079c80 left the live-region comment above toggleFlyoutScope instead of liveRegionMessage R16 Netero P3 Yes
CRF-202 P4 Author fixed (43743c8) FilterCombobox.test.tsx:2636 "commits a typed listed owner %s under the switch's key" no longer reaches the typed-Enter listedOption branch; disabling the branch passes 197 of 197 tests R16 Netero Yes
CRF-203 Nit Author fixed (43743c8) types.ts:66 scopeToggle doc line edited in 02631cd not rewrapped (99 columns) R16 Netero Yes
CRF-204 P4 Author fixed (43743c8) FilterCombobox.tsx:188 shownFlyoutKey comment says the scope flyout shows only when its row is highlighted or typing leaves no row highlighted; since 710a26d it also shows after the pointer leaves the menu R16 Razor P4, Gon P2 Yes
CRF-205 P4 Author fixed (43743c8) FilterCombobox.tsx:279 Comment above useFlyoutOptions describes click and Enter rules implemented elsewhere, after 710a26d removed the narrowed-text condition R16 Gon P2, Leorio P4 Yes
CRF-206 P2 Dropped by orchestrator ("as when" gives examples, not a complete list; Leorio, Meruem, Melody verified it matches every writer; round 15 asked for this update in CRF-193; CRF-174, CRF-186 precedent) FilterCombobox.tsx:180 highlightedCategoryKey comment lists writers of undefined, omits the initial state, and "another row" has no antecedent R16 Gon No
CRF-207 P4 Author fixed (PR description edited R16) useFilterCombobox.ts:1496 PR description says typed Enter keeps a held owner when no option is listed; code keeps it when no option is highlighted, which includes a listed owner after the pointer leaves the menu R16 Gon P4, Meruem P4, Leorio P4 Yes
CRF-208 Nit Author fixed (43743c8) useFilterCombobox.ts:506 scopeChipHolding comment uses "selected" (applied) and "selecting" (choosing) in one sentence R16 Gon Yes
CRF-209 Nit Author fixed (43743c8) FilterCombobox.test.tsx:3047 Two new test names say "shared" but type sha; "keeps the Owner flyout from" reads as "prevents" R16 Leorio, Gon Yes
CRF-210 Note Dropped by orchestrator (summary of Gon's other findings) FilterCombobox.tsx:279 Three comments describe code 710a26d changed R16 Gon No
CRF-211 OOS Out of scope (body) useFilterCombobox.ts:1499 In every submenu category, typed Enter on an applied value removes the chip when its option is highlighted and keeps it after the pointer leaves the menu (typed Enter never toggles; predates the PR) R16 Meruem No
CRF-212 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.tsx:598 CRF-193 fix verified and pinned; Escape and reopen keep the mirror in step with cmdk R16 Meruem No
CRF-213 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.tsx:598 CRF-193 fixed and pinned; CRF-192 condition removal cannot change the result R16 Razor No
CRF-214 Note Dropped by orchestrator (cleared concern, no change requested) types.ts:65 02631cd typed-Enter wording matches handleInputKeyDown R16 Razor No
CRF-215 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.test.tsx:2985 CRF-194 and CRF-195 fixed; 710a26d tests fail when their behavior is removed R16 Bisky No
CRF-216 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.tsx:180 highlightedCategoryKey comment accurate after 710a26d R16 Leorio No
CRF-217 Note Dropped by orchestrator (cleared concern, no change requested) 710a26d Commit subject names the condition and behavior R16 Leorio No
CRF-218 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.tsx:186 Typed sha flyout shows in every checked sequence and hides only while another row is highlighted R16 Nami No
CRF-219 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.tsx:598 Every highlightedCategoryKey writer matches cmdk's highlight R16 Melody No
CRF-220 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:1480 Typed Enter in Owner agrees with the base's shownOptions gating R16 Melody No
CRF-221 Nit Author fixed (2d06c84) FilterCombobox.tsx:189 CRF-204 fix left the scopeMatchShown comment line unwrapped at 126 columns R17 Netero Yes
CRF-222 P3 Dropped by orchestrator (disproved: removing the narrowed-text term from drillIn fails 2 tests, orchestrator mutation; the comment states the behavior at the code that implements it, as round 16 asked in CRF-205) FilterCombobox.tsx:602 drillIn comment restates categoriesNarrowedByText as the reason a click enters the category; claims no test pins the term R17 Gon No
CRF-223 P2 Dropped by orchestrator (round 16 asked for this sentence at this branch in CRF-205; it states the rule, and "keeps free text when Enter targets a row listed by the scope phrase" pins it; CRF-174 precedent) useFilterCombobox.ts:1585 Scope-row completion comment restates the condition and omits why the text is kept as a search R17 Gon No
CRF-224 P2 Dropped by orchestrator (round 16 asked to keep this sentence in CRF-205; it is accurate; same pattern as CRF-133, CRF-153) FilterCombobox.tsx:279 "The flyout renders only on wider viewports." reads !isMobile aloud R17 Gon No
CRF-225 Nit Dropped by orchestrator (the component names text matching searchPhrase a scope match: findScopeMatch, scopeMatchKey, scopeMatchedCategory) FilterCombobox.test.tsx:3236 Four test names call searchPhrase "the scope phrase" R17 Gon No
CRF-226 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.tsx:602 Round 16 text fixes (CRF-205, CRF-207, CRF-209) complete R17 Leorio No
CRF-227 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.test.tsx:2706 CRF-202 and CRF-204 fixes verified; moved comments match their code R17 Razor No
CRF-228 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.test.tsx:2707 CRF-202 fixed; test fails when the listed branch is removed R17 Bisky No
CRF-229 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.tsx:188 CRF-221 fixed; rewrapped comment still matches scopeMatchShown R18 Nami No
CRF-230 Note Dropped by orchestrator (cleared concern, no change requested) useFilterCombobox.ts:780 Base loop change to typeaheadOptionsByKey does not interact with the PR's categories wrapper R18 Nami No
CRF-231 Note Dropped by orchestrator (cleared concern, no change requested) FilterCombobox.tsx:1246 Merge 9141926 keeps the round 18 behavior for a failed flyout load and for the PR's activeOptions and typeaheadOptionsByKey readers R19 Melody No
CRF-232 P4 Author fixed (f8773ec) categoryOptions.tsx:136 Since 53e9abf the {username} (you) row disappears when the typed text matches the current user's name or email but not the username, while other users matched that way stay listed R20 Netero Yes
CRF-233 Nit Author fixed (f8773ec) WorkspacesFilter.stories.tsx:77 Comment added in 9f90a12 is 93 columns wide R20 Netero Yes
CRF-234 P4 Author fixed (f8773ec) WorkspacesFilter.stories.tsx:107 Story OrdinaryUserSeesOwnerAndUser and its comment describe the User category this PR removes R20 Leorio Yes
CRF-235 P4 Author fixed (f8773ec) categoryOptions.tsx:99 getSelfUserFilterOptions comment explains only the restricted-user caller; since 53e9abf it decides the self row for every user, and its match rule is written nowhere R20 Leorio P4, Zoro Nit Yes
CRF-236 OOS Out of scope (body) categoryOptions.tsx:101 getSelfUserFilterOptions repeats the filterOptionsByText match rule by hand (predates the PR) R20 Netero No
CRF-237 Note Dropped by orchestrator (cleared concern, no change requested) categoryOptions.tsx:136 Self row appears and disappears the same way in typed rows, flyout search, and the pending local filter R20 Nami No
CRF-238 Note Dropped by orchestrator (cleared concern, no change requested) categoryOptions.test.ts:37 53e9abf test fails when its behavior is removed R20 Bisky No
CRF-239 Note Dropped by orchestrator (cleared concern, no change requested) categoryOptions.tsx:136 Self-row match agrees with every Owner option consumer R20 Melody No
CRF-240 Note Dropped by orchestrator (cleared concern, no change requested) categoryOptions.tsx:136 Description's unknown-owner Enter claim matches the code R20 Razor No
CRF-241 Nit Author fixed (1142d5e) categoryOptions.test.ts:37 Test name "lists the current user only when the query matches it" is false since f8773ec lists the current user on a users API match R21 Gon, Leorio Yes
CRF-242 Nit Dropped by orchestrator (shadowing part disproved: a leftover [...self, ...options] fails tsc with TS2488 against the DOM global; the rest is a naming preference with no project rule) categoryOptions.tsx:128 self names an array of zero or one options and shadows the DOM global self R21 Gon No
CRF-243 P4 Dropped by orchestrator (disproved: breaking the lookup fails FilterCombobox.test.tsx:3203, orchestrator mutation; portaling the panel breaks cmdk itself, primitives.tsx:113) FilterCombobox.tsx:1022 FlyoutScopeToggle focuses the input through a cmdk DOM lookup instead of actions.focusInput R21 Ging React No
CRF-244 Note Dropped by orchestrator (cleared concern, no change requested) categoryOptions.tsx:126 New comment above self states the condition the code implements R21 Leorio No
CRF-245 Note Dropped by orchestrator (cleared concern, no change requested) categoryOptions.test.ts:44 CRF-232 fixed; the new test fails when the fix is removed R21 Bisky No
CRF-246 Note Dropped by orchestrator (cleared concern, no change requested) categoryOptions.tsx:128 CRF-232 fixed; the self row appears and hides like every other Owner row R21 Nami No
CRF-247 Note Dropped by orchestrator (cleared concern, no change requested) categoryOptions.test.ts:37 CRF-241 fixed; the new test name matches the fallback the test runs R22 Gon, Leorio, Nami, Bisky No

Law analysis

R5: effective +1227 -152 (13 files; 605 production, 622 test). Head 7ed2266. Verdict: Don't split. Enforcement: Advisory. Valid vertical cut exists (slice B: scope-phrase typeahead, about 80 production lines) but splitting at round 5 costs more review than it saves; the unreviewed base-merge focus change in primitives.tsx is handled as finding CRF-74.

R15: effective +1811 -129 (13 files; 634 production, 1177 test). Head 9a045a4. Verdict: Don't split. Enforcement: Advisory. Production grew 29 lines since R5; the rerun was triggered by test growth. Slice B (scope-phrase typeahead, about 100 production / 230 test) is still a valid vertical cut but depends on the scope toggle and Workspaces adoption, and splitting at round 15 would re-thread shownFlyoutKey and hasCategorySearchText through code both slices touch. CRF-74 (round 5 concern 6) is resolved. Full analysis: round-15/law.md.

Contested and acknowledged

CRF-8 (P2, categoryOptions.tsx:-161) - Author accepted R5

  • Finding: Removing the Shared attribute removes the only menu path to workspaces with any ACL entry (shared:true); the Owner switch cannot produce that set.
  • Author defense (PRRC_kwDOGkVX1s703Guz): The human author (@tracyjohnsonux) confirmed on the thread that losing the menu path to "any workspace with sharing" is accepted. The PR description now says the switch is the menu path to shared workspaces and that shared:true still works when typed.
  • Author accepted: This is the human decision the R3 panel re-raise (Pariston, Luffy, 0/2 accept of the earlier defense) asked for. The fact the finding stated is no longer disputed; the gap is accepted by a human, not by an agent. Consequence that remains: DisableWorkspaceSharingDialog.tsx:41 counts a set the menu cannot list, and typed shared:true renders as free text (CRF-7).

CRF-12 (P3, useFilterCombobox.ts:288) - Acknowledged R4

  • Finding: With the switch on by default, picking an owner commits user:<name> and includes workspaces owned by others; batch actions on the filtered list act on those rows.
  • Author defense (PRRC_kwDOGkVX1s70uB2C): Author decision: the switch stays on by default when an owner is picked; the pill next to the chip shows the widened scope, and removing it narrows the chip.
  • Author accepted: Recorded as an author decision after Workspaces filter: decide Owner scope toggle defaults and wording #29912 item 2. The reply does not address the batch-action consequence. With 886074f the pill reads + shared with <owner>, which states the widening next to the chip.

CRF-8 (P2, categoryOptions.tsx:-161) - Removing the Shared attribute drops the any-sharing filter

  • Finding: shared:true (any ACL entry) was the only menu path to workspaces that have any sharing; the Owner switch only chooses owner:X or user:X and cannot produce that set. Proposed: keep the Shared attribute, relabel it if misread.
  • Author defense (PRRC_kwDOGkVX1s70fPyg): Removal is intentional; the team dogfooded the Owner scope toggle as the replacement, "a different way to reach shared workspaces, not a removed capability"; existing shared:true URLs still apply on the backend.
  • Panel re-raise (R3, 0/2 accept): Pariston and Luffy, the agents that raised it in round 1 (reused, not new), re-raised. Both verified no code change and that no switch state returns the any-ACL set. New evidence: DisableWorkspaceSharingDialog.tsx:41 counts organization:<id> shared:true and warns the org admin how many workspaces will be unshared, but after this PR that list is reachable only by typing shared:true, which renders as free text (CRF-7). Both state that removal may stand only as an explicit human decision made knowing the capability is lost.

CRF-10 (P3, useFilterCombobox.ts:272) - No-chip switch state depends on removal history

CRF-12 (P3, useFilterCombobox.ts:288) - Switch on by default

  • Finding: With no chip the switch is on, so picking an owner commits user:<name> and includes workspaces owned by others; batch actions on the filtered list act on those rows. Proposed: default off when no chip exists.
  • Author defense: Deferred for the human author's decision (reply PRRC_kwDOGkVX1s70saKZ).
  • Do not re-evaluate. Tracked in Workspaces filter: decide Owner scope toggle defaults and wording #29912 (item 2).

CRF-13 (P3, WorkspacesFilter.tsx:69) - Pill text reads as a narrowing condition

  • Finding: shared with owner next to AND chips reads as "my workspaces that are shared"; the remove button label does not say it drops shared workspaces. Proposed: a label stating the OR, such as or shared.
  • Author defense: Deferred for the human author's decision (reply PRRC_kwDOGkVX1s70saQ4).
  • Do not re-evaluate. Tracked in Workspaces filter: decide Owner scope toggle defaults and wording #29912 (item 3).

CRF-14 (P3, WorkspacesFilter.tsx:67) - No-chip switch label

  • Finding: With no Owner chip, "Include shared workspaces" promises a list change, but flipping the switch emits no query. Proposed: a label saying it applies to the selected owner.
  • Author defense: Deferred for the human author's decision (reply PRRC_kwDOGkVX1s70saWp).
  • Do not re-evaluate. Tracked in Workspaces filter: decide Owner scope toggle defaults and wording #29912 (item 4).

CRF-15 (P3, FilterCombobox.tsx:408) - Truncated pill text on touch and keyboard

  • Finding: On phones the pill truncates and the full text is only in a Radix tooltip on a non-focusable span, which ignores touch. Proposed: a label short enough not to truncate, or the full text in a reachable place.
  • Author defense: Deferred for the human author's decision (reply PRRC_kwDOGkVX1s70saa8).
  • Do not re-evaluate. Tracked in Workspaces filter: decide Owner scope toggle defaults and wording #29912 (item 5).

CRF-72 (P3, docs/user-guides/shared-workspaces.md:85) - User guide still says to filter for "Shared"

Round log

Round 1

Full panel (round 1). Netero first pass: 2 P2, 2 P3, 1 P4, 1 Nit, 1 Note; no P0, so the panel ran. Law not spawned (effective additions 588, under 1000).

Panel: ging-ts, ging-react, nami, luffy, melody (domain specialists), pariston, mafuuu (core), bisky (tests changed), gon, leorio (once-per-PR floor, round 1 placement), meruem (wildcard, random draw from the distilled set). Size target for a focused TypeScript PR is 6 to 8; the five triggered specialists plus core and Bisky reach 8, and Gon and Leorio were added for their round 1 floor.

Cross-check:

  • CRF-30 disproved empirically. Keep-argument: a repeated user key is rejected by parseSingle, which would be P1. Scratch test (initialValue: "user:bob", pick alice) emitted ["user:bob"], ["user:alice"]; updateFromChips passes tokens through composeFilterQuery -> dedupeChips -> dedupeInOrder, which keeps one value per key. CRF-44 falls with it.
  • CRF-9 verified: scratch test with owner:alice user:bob rendered two Owner chips plus the pill, and one Backspace emitted owner:bob. CRF-2 verified: typing owner:alice{Enter} emitted user:alice.
  • CRF-31 dropped. Keep-argument: read in isolation, withOptionToken returns two same-key chips, contrary to its comment. Drop reason: the emitted query holds one chip for pick paths; the case where the comment is false (mixed keys from URL or typing) is CRF-9.
  • CRF-24, CRF-25, CRF-26 downgraded to Nit. Keep-argument: Gon rated them P2/P3 because comments that restate or misattribute code mislead maintainers. Downgrade reason: the code works and the consequence is reading cost only; the severity rules classify comment wording, naming, and unnamed constants as Nits. Leorio rated CRF-25 Nit.
  • CRF-10 merges five reviewers on narrowedScopes; Leorio's comment inaccuracy at line 270 is the same defect. CRF-11 merges Luffy's removal proposal with Meruem's Enter reproduction. CRF-9 merges Mafuuu's pill-placement case.
  • Interactions: CRF-11's removal would also remove the typedFilterText clearing behind CRF-1 and CRF-23. CRF-8 makes CRF-7 moot. CRF-12 and CRF-13 compound: the chip reads owner: alice while the query includes workspaces owned by others.

Reviewed against 5c717b1..74472cd.

Round 2

BLOCKED. Churn guard: 21 addressed, 1 contested (CRF-8, author reply PRRC_kwDOGkVX1s70fPyg), 5 silent (CRF-10, CRF-12, CRF-13, CRF-14, CRF-15). The author's follow-up comment IC_kwDOGkVX1s8AAAABW7ECMA lists the five as "Still open for an author decision" with no reasoning or ticket, which the churn guard rules classify as silent. No Netero, no panel. Addressed statuses record the author's claimed fix commits; the panel has not verified them. The PR base moved to 19fe0bf (stacked on #29876); effective additions grew 588 to 1000, below the Law threshold. The same follow-up comment answers the round 1 out-of-scope items: CRF-40 belongs to the base redesign (#29426), CRF-41 (categoryPreview) no longer exists on this branch, CRF-42 needs a backend change, CRF-43 predates the PR.

Reviewed against 19fe0bf..ab755ae.

Round 3

Churn guard: PROCEED. CRF-8 contested; CRF-10, CRF-12 to CRF-15 deferred with #29912 items 1 to 5 (issue verified open by the churn guard). Head unchanged since round 2 (ab755ae), but round 2 ran no reviewers, so Netero ran: 2 P3, 4 P4, 1 Nit (advisory, post-panel round). Law not spawned (effective additions 1000, not above 1000; growth since round 1 is 412).

Restructure round: files added to the diff since the last panel round (filterQuery.test.ts, primitives.tsx, WorkspacesFilter.test.tsx), and an exported field was renamed (scopeToggle.chipKey to widenedKey) with a new export (categoryChipKeys). Full panel: ging-ts, ging-react, nami, luffy, melody (specialists), pariston, mafuuu (core), bisky (tests), gon, leorio (restructure round), knov (wildcard, random draw). Pariston and Luffy are the CRF-8 dispute check-in and were reached through their round 1 agents with message_agent; they do not count toward the size target of 8.

Cross-check:

  • CRF-8: 0/2 accept, re-raised with new evidence (DisableWorkspaceSharingDialog.tsx:41). Posted as a reply on its thread.
  • CRF-52: four reviewers reproduced that scopeGroupKey drops one of owner:/user: while WorkspacesPage.tsx:116 sends the raw URL query (verified by reading both). The collapse is the fix round 1 proposed for CRF-9, so that proposal was wrong; the comment says so. Mafuuu and Melody proposed emitting the collapsed query on reconcile; Pariston and Knov proposed keeping the other token. Chose keeping the token as free text: emitting on load changes the results without user action. Severity P2 (Mafuuu, Knov) over P3 (Pariston, Melody): Mafuuu showed the shape user:me owner:alice comes from main's default query plus one Owner pick. Posted as a new finding on the CRF-9 thread.
  • CRF-53: new problem from the CRF-2 fix; posted on the CRF-2 thread. Interacts with CRF-46: applying the typed key on pick removes the duplicated block.
  • CRF-45, CRF-56 assume the one-chip rule survives the CRF-52 fix; the free-text fix keeps it, a two-chip fix would invalidate them.
  • CRF-18: Bisky verified the suggestion half fixed and the typed-Enter half still untested (mutation at line 1125 passes); re-raised on its thread without a new ID.
  • CRF-60, CRF-61 downgraded from Gon's P2 to Nit. Keep-argument: duplicated rule comments drifted in round 1 (CRF-22). Downgrade reason: code works, cost is maintenance only, the severity rules class comment wording as Nit, and Leorio rated CRF-60 Nit. Two of the four copies in CRF-60 were requested by round 1 (CRF-20, CRF-24), so the ask keeps those two.
  • CRF-66 dropped. Keep-argument: chipKey next to chipKeys reads as element and list. Drop reason: two of the three fields predate the PR, round 1 (CRF-21) told the author these carry the current-key meaning, and the one new field follows them; the pre-existing names are listed out of scope (CRF-67).
  • CRF-65 moved in scope from Gon's out-of-scope list: the PR changed what the aliases example depends on (removed the aliases: ["user"] uses, made user the widenedKey).
  • Verified fixed by reviewers: CRF-11, CRF-19, CRF-28 (Luffy), CRF-19 (Nami), CRF-17 (Bisky).

Post result: 17 inline comments (filterQuery.ts:52 for CRF-58 and types.ts:24 for CRF-65 folded into the body) and 4 replies. All three unresolve actions failed ("Resource not accessible by integration"), so the replies carrying CRF-52 (CRF-9 thread), CRF-53 (CRF-2 thread), and the CRF-18 re-raise sit in resolved, collapsed threads. The round 3 body names CRF-52 only indirectly, the CRF-18 re-raise briefly, and CRF-53 not at all, contrary to the vault rule in projects/my-agent/deep-review-resolved-thread-invisible.md. Round 4: if the author has not responded to CRF-52, CRF-53, or the CRF-18 re-raise, treat that as not seen rather than silent, restate all three in the body with the response needed, and post new findings inline instead of as replies on resolved threads.

Reviewed against 19fe0bf..ab755ae.

Round 4

BLOCKED. Churn guard: 4 addressed (CRF-10, CRF-13, CRF-14, CRF-15 in 886074f), 1 acknowledged (CRF-12, author decision to keep the default on; the reply does not engage the batch-action consequence, and no panel has evaluated it), 23 silent (CRF-8 re-raise, CRF-18 re-raise, CRF-45 to CRF-65). No Netero, no panel.

Timing: 886074f was committed at 14:45:17Z, before the round 3 review posted at 14:47:29Z; the author's replies (14:50:27Z to 14:50:39Z) answer the #29912 items, and /coder-agents-review followed at 14:50:44Z. The author-agent finished its #29912 work in parallel with round 3 and requested a review three minutes after round 3 posted, so the silence on round 3 comes from timing, not refusal. The rules still classify it silent.

The round 3 log asked to treat CRF-52, CRF-53, and the CRF-18 re-raise as not seen; the churn guard has no such status and classified them silent, and the verdict is BLOCKED without them. The round 4 body restates all three in full, plus CRF-58 and CRF-65, which have no thread because round 3 folded them into its body.

Reviewed against 19fe0bf..886074f.

Round 5

Churn guard: PROCEED. 22 addressed in 7ed2266 (CRF-18, CRF-45 to CRF-65), CRF-8 accepted by the human author on its thread, zero silent. Netero ran (post-panel round): 1 P2, 1 P3, 1 Note. Law ran (effective additions 1227, above 1000): Don't split, Advisory, 1 P3 (CRF-74).

Full panel: the base moved to d15c0c7 (merge of #29876 updates, including the primitives.tsx focus change) and effective additions grew from 1000 to 1227 since the last panel round. Panel: ging-ts, ging-react, nami, luffy, melody (specialists), pariston, mafuuu (core), bisky (tests), gon, leorio (naming and docs changed), knov (wildcard, random draw from the distilled set). Ging TS and Ging React: no findings.

Cross-check:

  • CRF-71 and CRF-75 are two failures from one cause: the CRF-52 fix keeps the second owner:/user: token in the editable input, and neither the input parser nor toggleScope accounts for it. Verified by reading useFilterCombobox.ts:997-1028 (parseTypedCategoryPrefix matches user: inside the kept text, commitTextBeforePrefix emits only the text before it) and :749-768 (rewrite of chipValues plus extractFreeText of the last query). Netero, Pariston, Mafuuu, and Knov each reproduced with scratch tests; no orchestrator scratch test. Kept as two entries: different actions, different consequences (filter silently dropped vs. rejected query), different fixes. Contradiction on the fix: Netero and Pariston move the token out of the input (plain chip or separate state); Mafuuu and Knov block the rewrite when the kept token has the target key. Moving the token out does not stop the repeated key on a switch flip, so CRF-71 carries the move and CRF-75 carries the guard. CRF-52 status unchanged: its load-time drop is fixed; the author's claim that "edits do not widen the results" (IC_kwDOGkVX1s8AAAABXAIMrA) is false for input edits (CRF-71).
  • CRF-75 severity P2, not P1: one click shows a validation error, recoverable by flipping back or clearing the text; no data change. Line 765 (Mafuuu, Knov) over 761 (Pariston): it is the free-text append that causes the repeat.
  • CRF-73 posted as Note (body): the repeated same key is already rejected on load before and after the PR; the change is that the first edit no longer repairs it, and that edit then hits CRF-71.
  • CRF-77 interacts with CRF-71: if the kept token moves out of extractFreeText, the optional categories default stops mattering for it.
  • CRF-78 interacts with CRF-75: the guard CRF-75 asks for should share one predicate with the switch disabled rule.
  • CRF-76 verified by reading useFilterCombobox.ts:338-355 (isScopeWidened returns the typed key in category mode) and FilterCombobox.tsx:400-403; the removal case follows from toggleScope rewriting only fromKey tokens. CRF-81's comment ("The scope pill follows its category's chip") becomes true once CRF-76 is fixed.
  • CRF-79: Luffy P4, three others P3; higher wins. Verified the doc (types.ts:54-57) and the PR description ("Each pick turns it on").
  • CRF-80 merges Gon and Leorio's two stale "pill label" comments (hook and story) into one entry; one rename left both.
  • CRF-81, CRF-82 downgraded from Gon's P2 to Nit. Keep-argument: AGENTS.md requires substantive comments and the local name shadows the category field with a different meaning. Downgrade reason: code works, cost is reading only, the severity rules class naming and comment wording as Nit.
  • CRF-83 dropped. Keep-argument: "Shared by" implies several consumers, and there is one caller. Drop reason: the comment says the Owner category uses these options for both owner and user tokens, which is true and explains why the options carry no key.
  • CRF-72 in scope: the diff removes the Shared option and so makes docs/user-guides/shared-workspaces.md:85 false. Line is outside the diff; post folds it into the body.
  • CRF-74 verified: git show f094849413 deletes remove() and the test "removes a chip with %s on its remove button and focuses the input"; f094849 is in the base; the effective diff re-adds remove() for every chip.
  • CRF-8 and CRF-12: Pariston and Luffy, who raised them, did not re-raise.
  • Pariston's out-of-scope item (no OR in the query language) is CRF-42; not relisted.

Placement: CRF-71 and CRF-75 are posted inline on their own lines, not as replies on the resolved CRF-52 thread (vault: resolved threads hide replies); the body names them as new evidence against the mixed-query fix.

Post result: review 5322464645 (COMMENT), 11 inline comments; CRF-72 (docs/user-guides/shared-workspaces.md:85) folded into the body. No replies, no thread actions.

Reviewed against d15c0c7..7ed2266.

Round 6

BLOCKED. Churn guard: 11 addressed in 0ea86e5 (CRF-71, CRF-74 to CRF-82, CRF-84), 1 silent (CRF-72). No Netero, no Law, no panel. The addressed statuses record the author's claimed fixes; no reviewer has verified them.

CRF-72: the author comment IC_kwDOGkVX1s8AAAABXBJecQ defers the docs/user-guides/shared-workspaces.md update until the feature lands in dogfood, at the PR author's request, with no linked issue. A deferral without a ticket is classified silent. The comment does not say what users of the published guide see in the meantime. The body asks for a linked issue, a fix in this PR, or an explicit human acceptance on the PR.

CRF-73 (Note (body)): the same comment says one chip per key is restored, so a repeated owner:a owner:b keeps the last value as in the base. Not classified; the panel verifies it with the rest of 0ea86e5.

0ea86e5 replaces the kept-search-text approach with a second plain chip, removes the categories parameters from four exported filterQuery.ts helpers, and adds isScopeToggleDisabled, scopePillCategoryKey, and removeScopePill. The next unblocked round is a restructure round (exported signatures changed) and gets a full panel. The base moved to 0b1bba8; effective additions are 1294, up 67 from Law's last analysis at 1227, so Law does not rerun.

Post result: review 5322791094 (COMMENT), body only.

Reviewed against 0b1bba8..81773d0.

Round 7

Churn guard: PROCEED. CRF-72 deferred to #29955 (open, names line 85 and the screenshot; checked through the public API because gh returned 401). Head f33837c: 81773d0..f33837c only merges #29876. Netero ran (post-panel round): 3 P4 (CRF-90 to CRF-92). Law not spawned: effective additions 1294, growth 67 since round 5's 1227.

Restructure round: 0ea86e5 (unreviewed, round 6 was blocked) removed the categories parameter from four exported filterQuery.ts helpers and replaced the kept-search-text design with a second plain chip. Full panel: ging-ts, ging-react, nami, luffy, melody (specialists), pariston, mafuuu (core), bisky (tests changed), gon, leorio (restructure round), zoro (wildcard, random draw from the distilled set). Size 11 exceeds the 6 to 8 target because Gon and Leorio are mandatory on a restructure round. No contested findings, so no dispute check-in.

Cross-check:

  • Prior fixes verified by reviewers: CRF-71, CRF-75 (Pariston, Mafuuu), CRF-74, CRF-77 (Netero), CRF-77, CRF-78 (Bisky), CRF-76 (Melody), CRF-79 to CRF-82 (Gon), CRF-79 (Luffy). CRF-73 obsolete: dedupeInOrder keeps the last value per key again (Netero). Ging TS, Ging React, Pariston: no findings.
  • CRF-93 verified by reading withCategoryOption (useFilterCombobox.ts:818-830): it keeps the scope chip, drops every later chip under the picked key, then replaces the scope chip, so owner:bob user:carol plus a typed user: pick gives user:alice. The code comment and the test at FilterCombobox.test.tsx:1862 state this as intended; Nami's alternative (replace the chip that already has the picked key) keeps the untouched filter.
  • CRF-94 merges Luffy's P3 (two owner: chips, non-removable pill; render raw keys, no pill) with Nami's and Leorio's P4 (no reason for the disabled state; add a reason line). Higher wins: P3. The display with the pill is logically equivalent to the query ((me or shared with me) and owner carol), so the finding is framed as "cannot tell what blocks the switch or how to change it", not as a misreported query. Luffy's rendering removes the pill state; the reason line is the smaller fix. Both are offered.
  • CRF-95 in scope: hasCategorySearchText and the clearing branches are PR code (Law round 5, concern 3). Trigger is narrow (applied search that is a 3+ character prefix of shared with owner); kept at Melody's P3, no reviewer disagreed. Melody's out-of-scope item (selectCategory at :751 drops an applied search) is the CRF-40 class already listed; not relisted.
  • CRF-96: Melody and Zoro independently. CRF-102 (Zoro's boolean typedScope) also fixes it.
  • CRF-92 now includes Zoro's single-accessor proposal; Netero's text proposes the same accessor, so they are one finding.
  • CRF-99: Bisky's mutation shows the guard prevents owner:carol (the user:me filter silently dropped by per-key dedupe), not a rejected query; the comment at useFilterCombobox.ts:342 and the PR description give the old reason.
  • CRF-104 downgraded from Gon's P3 to Nit. Keep-argument: a reader at FilterCombobox.tsx:400 can read isPlainScopeChip(token) as "is the scope chip". Downgrade reason: "plain" marks it as not the scope chip, matching the PR description's "plain chip"; code works; naming is Nit.
  • CRF-105 downgraded from Gon's P2 to Nit, same reason as CRF-82 in round 5; the two comments are merged into one finding.
  • CRF-98 follows CRF-79's fix; posted inline as a new finding because the CRF-79 thread is resolved.
  • Interactions: CRF-103's shorter toggleCategoryOption keeps the hasCategorySearchText ? "" : typedFreeText expression that CRF-95 changes. CRF-101 and CRF-107 are on the same story.

Post result: review 5323286244 (COMMENT), 18 inline comments, no replies.

Reviewed against 4cb81dc..f33837c.

Round 8

Churn guard: PROCEED. 18 addressed in 64dbfce (CRF-90 to CRF-107), CRF-72 deferred to #29955. Head 64dbfce; base moved to d592615. Netero ran (post-panel round): 1 P4 (CRF-112). Law not spawned: effective additions 1344, growth 117 since round 5.

Fix round: 64dbfce only fixes round 7 findings. The hook's return shape changed (scopeState replaces three accessors, showsOwnQueryKey replaces isPlainScopeChip), but that is the change CRF-92 and CRF-104 asked for, and the skill says a round that only fixes prior findings is not a restructure. Panel: personas that raised open findings the commit touches (nami, luffy, leorio, melody, zoro, mafuuu, bisky, gon), plus ging-ts and ging-react for the changed .ts/.tsx files, plus razor (wildcard, random draw from the distilled set excluding mafuuu and zoro). No contested findings.

Cross-check:

  • Prior fixes verified by reviewers: CRF-93, CRF-94 (Nami, Luffy), CRF-95, CRF-96 (Melody), CRF-97, CRF-98 (Mafuuu), CRF-99 to CRF-101 (Bisky), CRF-92, CRF-96, CRF-102, CRF-103 (Zoro), CRF-104 to CRF-107 (Gon), CRF-91 to CRF-106 (Netero, by reading and mutation). Ging TS, Ging React, Luffy, Zoro: no findings.
  • CRF-113: three reviewers reproduced it independently (user:me owner:carol plus a click on carol emits user:carol owner:carol; owner:bob user:carol plus carol emits owner:carol user:carol). Verified by reading: optionChipKey (useFilterCombobox.ts:386) follows the first chip, the panels mark a row selected only for optionToken(chipKey, option), and withCategoryOption replaces the chip under the picked key. Mafuuu P3 over Melody and Nami P4: a click drops a filter and the list cannot remove the second chip's owner.
  • CRF-114 merges Gon (types.ts:63) and Leorio (types.ts:61, reproduced user:bob plus typed creator: gives owner:alice); Leorio's reproduction is the sharper evidence. The alias-narrows behavior is what CRF-96 asked for; only the doc lags.
  • CRF-115 merges Bisky's mutation (alias set to null passes) with Leorio's rename.
  • CRF-117 in scope: the PR changed the comment's second line and added the listedOnlyByScopeMatch Enter branch that makes the first line false.
  • CRF-119 to CRF-121 dropped. Keep-arguments: each comment opens with a clause that restates the code below it, the pattern CRF-105 removed. Drop reasons in the rows: CRF-120's premise is wrong (the consequence is stated); CRF-121's comment already carries the part Gon wants; CRF-119's comment gives the state's role and override, and its alias gap goes into CRF-114. CRF-118 kept as Nit because it matches CRF-105, which the author accepted.
  • CRF-123 kept out of scope as Gon filed it: the aliases doc line is unchanged; the PR's new reliance on no-colon alias matching does not change what the doc gets wrong.

Post result: review 5323550583 (COMMENT), 7 inline comments, no replies.

Reviewed against d592615..64dbfce.

Round 9

Churn guard: PROCEED. 7 addressed in 5dab005 (CRF-112 to CRF-118), CRF-72 deferred to #29955 (open). Head 5dab005; base moved to 6353285. Netero ran (post-panel round): 1 P3, 1 P4 (CRF-127, CRF-128). Law not spawned: effective additions 1412, growth 185 since round 5. The author answered the round 8 out-of-scope items: CRF-122 predates the PR and carries TODO(DEVEX-421 follow-up); CRF-123 fixed in 5dab005 (aliases doc).

Fix round: 5dab005 only fixes round 8 findings (+163 -76 in 4 files). Panel: personas that raised open findings the commit touches (mafuuu, melody, nami for CRF-113; leorio, gon for CRF-114, CRF-116, CRF-118; bisky for CRF-115; razor for CRF-117), ging-ts and ging-react for the changed .ts/.tsx files, chopper (wildcard, random draw from the distilled set excluding mafuuu and razor). No contested findings.

Cross-check:

  • Prior fixes verified by reviewers: CRF-113 option panels (Nami by scratch test, Mafuuu, Melody), CRF-114 to CRF-117 (Leorio, Bisky, Razor), CRF-118 (Gon), CRF-112 (Bisky). Ging TS, Ging React, Nami: no findings.
  • CRF-127 merges Netero's value-suggestion path (:694) with Melody's and Razor's typed-Enter path (:1251): the same cause (both still build the token with optionChipKey), the same fix (categoryOptionToken), so one finding. Melody P3 and Netero P3 over Razor P4. Razor's alternative (a no-op in withCategoryOption when a scope chip holds the value) does not reach value suggestions, which commit through toggledChips; the comment gives Melody's fix.
  • CRF-129: Chopper reproduced the backend half with a coderd test (member q=user:<alice> 400, q=owner:<alice> 200). Verified by reading: parseUser (coderd/searchquery/search.go:712-731) looks up the username through the authorized store, and the member role reads only its own user (coderd/rbac/roles.go:460). The frontend half by reading: with the default user:me, optionChipKey is user, so the typed Enter candidate at :1251 is user:alice. Members without viewDeploymentConfig see only the "me" option, so typed Enter is their only menu path to another owner. Not a re-raise of CRF-12 (default on): the consequence is a validation error for members, not a widened result, and the fix changes only unmatched typed values. Kept at P2: an error instead of results, recoverable by flipping the switch off. The backend root cause is out of scope (CRF-135). Interaction: CRF-127 and CRF-129 change the same candidate line; categoryOptionToken for the two-key state and the category key for unmatched typed values combine.
  • CRF-130 merges Bisky's flyout test gap and Mafuuu's Enter-highlight test gap on the same fix. Bisky's note that the highlight check is reachable only after a typed prefix is not accepted: after typing, ArrowDown highlights a row, and the check then decides between toggling that row and committing the typed text.
  • CRF-133 and CRF-134 dropped (reasons in the rows). CRF-133's keep-argument: the sentence documents a branch in another file and leaves out Tab. CRF-134's keep-argument: its first clause restates the expression, the CRF-105 pattern.
  • Out of scope: CRF-135 (backend user: lookup) and CRF-136 (aria-hidden check mark, present at base FilterCombobox.tsx:586).

Post result: review 5323734896 (COMMENT), 6 inline comments, no replies.

Reviewed against 6353285..5dab005.

Round 10

Churn guard: PROCEED. CRF-127 to CRF-132 addressed in 92c95ec, CRF-72 deferred to #29955, none silent. Head ce9e642; base 200e857. Netero ran (post-panel round): 1 P3, 2 P4 (CRF-140 to CRF-142). Law not spawned: effective additions 1519, growth 292 since round 5.

Fix round: 92c95ec only fixes round 9 findings (+143 -37 in 4 files); the other new commits merge the #29876 base. Panel: personas that raised open findings the commit touches (melody, razor for CRF-127; chopper for CRF-129; bisky, mafuuu for CRF-130; leorio for CRF-131; gon for CRF-132), ging-ts and ging-react for the changed .ts/.tsx files, nami for UI state, zoro (wildcard, random draw from the distilled set excluding mafuuu, chopper, razor). No contested findings.

Cross-check:

  • Prior fixes verified by reviewers: CRF-127, CRF-128, CRF-130 (Netero mutation of :1298), CRF-131 (Leorio), CRF-132 (Gon), CRF-129 for owners missing from the options (Nami, Chopper). Chopper confirmed searchquery: user: filter rejects usernames the requester cannot read while owner: accepts them #29961 is open and covers the backend user: rejection (CRF-135). Ging TS, Ging React, Nami, Bisky, Chopper: no new findings.
  • CRF-140 verified by orchestrator scratch test (removed): Owner drill-in, type alice, wait for the alice row (data-value="user:alice", aria-selected="false"), Enter. From user:alice and from user:me it emitted owner:alice; from owner:me user:alice it emitted the query unchanged. In category mode autoHighlight is on, but the highlight stays on the category row (data-value="owner"), so the typed-Enter branch runs. Not checked in a real browser. Kept at P3: Enter on a listed owner silently drops shared workspaces that a click on the same row includes.
  • CRF-143 merges Leorio P3 (PR description), Razor P4 (same behavior, prefers fixing the description because typed Enter on an applied value is a no-op in every category: commitCategoryOption never toggles, verified at :922-928), and Gon P4 (appliedScopeChipFor comment at :393 names only click removal). One behavior decision settles all three texts. Kept at P3 (Leorio) over P4: keep-argument, the description is the PR's statement of behavior and becomes the squash commit body, and a reviewer who follows it sees Enter do nothing. Distinct from CRF-141, which is the missing test for the same expression.
  • CRF-144 merges Mafuuu P4, Zoro P4, Razor Nit (no searchquery: user: filter rejects usernames the requester cannot read while owner: accepts them #29961 link) and Leorio P4 (Owner wording in the generic hook). Leorio's point that the comment omits the applied-chip branch is folded into CRF-143. Razor Nit raised to P4 by the tiebreak.
  • CRF-145 (Zoro): equivalence checked by reading; with typedScopeWidened === true both forms give chipToken(widenedKey, value), with no scopeToggle both give chipToken(key, value). Interaction: the CRF-140 fix adds one condition to the key choice.
  • CRF-146 merges Gon Nit and Leorio Nit; the comment gives Gon's name optionTokenFor because the hook and the panel props already use it. Leorio's tokenFor lost on one-name-per-thing.
  • CRF-148 and CRF-149 dropped. CRF-148 keep-argument: it confirms the CRF-127 fix reaches every token producer. CRF-149 keep-argument: it records why the chipKey field became a callback. Both ask for no change, and the body does not list cleared concerns.
  • Out of scope: none new.

Post result: review 5324028773 (COMMENT), 8 inline comments, no replies.

Reviewed against 5dab005..ce9e642.

Round 11

Churn guard: PROCEED. CRF-140 to CRF-147 addressed in 3302857, CRF-72 deferred to #29955, none silent. Head 3302857; base 4083105. Netero ran (post-panel round): 1 P4 (CRF-150). Law not spawned: effective additions 1579, growth 352 since round 5.

Fix round: 3302857 only fixes round 10 findings (+89 -29 in 4 files); the other new commits merge the #29876 base, and the merge resolutions only list the base's inlineSections next to the PR's return fields (Netero, Nami, Zoro compared). Panel: personas that raised open findings the commit touches (leorio, razor, gon for CRF-143; mafuuu, zoro, leorio, razor for CRF-144; zoro for CRF-145; gon, leorio for CRF-146; gon for CRF-147), ging-ts and ging-react for the changed .ts/.tsx files, nami for UI state, melody for the token producer and consumer pairing, bisky for the new tests, knov (wildcard, random draw from the distilled set excluding mafuuu, zoro, razor). No contested findings.

Cross-check:

  • Prior fixes verified: CRF-140 and CRF-141 by mutation (Netero, Nami, Mafuuu, Zoro, Bisky, Melody, Razor), CRF-142 to CRF-147 by reading (Netero, Leorio, Gon, Zoro, Razor). Ging TS, Ging React, Nami, Melody, Zoro, Knov: no findings. Nami, Melody, Knov checked that typing one's own username highlights the self row, so Enter goes through that row and not the typed-Enter branch.
  • CRF-150 (Netero): verified by reading. listed uses option.value === typedOption.value, while optionMatches (filterQuery.ts:277) lowercases both sides, so a differently cased name shows the row but is not listed. Kept at P4: same consequence as CRF-140 (Enter narrows while the switch is on), reached only by typing a different case.
  • CRF-151 merges Leorio P4 and Bisky P4 on the same test. Bisky's mutation (toggleCategoryOption in the typed-Enter branch) passes all 163 tests. Distinct from CRF-141, which asked for a test of the lookup; the new test pins the lookup (Netero mutation) but not the kept chip.
  • CRF-152 merges Razor, Gon, Mafuuu (types.ts:63 omits the applied-chip exception; Mafuuu adds the hook comment at :1353, Gon the PR description sentence) and Leorio's PR description P4 ("highlighting" removes a chip, but only highlight plus Enter does). Leorio's sentence is the wording round 10 suggested in CRF-143; it is folded into this finding because the author edits the same description for the applied-chip omission and Leorio gave no code line. Precedent for doc plus description in one comment: CRF-79.
  • CRF-153 (Gon P2) dropped. Keep-argument: the comment is a second copy of the types.ts rule, and within one commit that copy already omits the applied-chip case, the drift pattern of CRF-22 and CRF-60. Drop reason: round 10 asked for the rule in this comment (CRF-144), the drift is posted as CRF-152, and CRF-152's fix offers Gon's shorter comment. Same pattern as CRF-133.
  • CRF-154 to CRF-158 dropped as cleared concerns that ask for no change. Keep-argument for each: they record verification of the round 10 fixes and the listed invariant (Knov: listed owners are users the requester can read). The body does not list cleared concerns.
  • Out of scope: none new.

Post result: review 5324223263 (COMMENT), 3 inline comments, no replies.

Reviewed against ce9e642..3302857.

Round 12

Churn guard: PROCEED. CRF-150 to CRF-152 addressed in cee3c66, CRF-72 deferred to #29955, none silent. Head cee3c66; base b25c4ac. Netero ran (post-panel round): 1 P4 (CRF-159). Law not spawned: effective additions 1586, growth 359 since round 5.

Fix round: cee3c66 only fixes round 11 findings (+42 -35 in 3 files); the other new commits merge the #29876 base, and the merge resolutions keep both sides (the base's flyoutLoadFailed and categoryLoadErrorMessage next to the PR's scope props; read by the orchestrator in aca623b). Panel: personas that raised open findings the commit touches (leorio, bisky for CRF-151; razor, gon, mafuuu, leorio for CRF-152), ging-ts and ging-react for the changed .ts/.tsx files, nami for UI state, melody for the token producer and consumer pairing, meruem (wildcard, random draw from the distilled set excluding mafuuu and razor). No contested findings.

Cross-check:

  • Prior fixes verified: CRF-150 for typed text (Netero mutation), CRF-151 (Netero, Bisky mutation, Leorio). CRF-152's two-key part holds; its wording is overbroad (CRF-161). Ging TS, Ging React, Nami, Melody: no findings. Melody confirms CRF-159 is the one token pair that still disagrees on case.
  • CRF-159 (Netero): verified by reading. optionTokenFor(activeCategory, listedOption) passes the option's spelling to appliedScopeChipFor, which compares with ===, so user:Alice no longer matches after cee3c66; before it, the typed Alice matched the chip exactly. Fix-chain regression for typed Enter; the click path had the same gap since the CRF-113 fix. Kept at P4: silent loss of owner:me, reached only through a chip spelled in another case.
  • CRF-160 (Bisky) is distinct from CRF-159: same-case applied owner on the listed branch, a test gap. Interaction: Bisky's single lookup (appliedScopeChipFor(activeCategory, listedOption ?? typedOption)) is the natural place for CRF-159's case-insensitive compare and for CRF-161 option one.
  • CRF-161 merges Razor P4 and Mafuuu P4, both reproduced with initialValue: "user:zed": Enter emits owner:zed. The overbroad sentence is the wording round 11 suggested in CRF-152 (journal entry 2026-09-26 on suggested text recurs). The comment gives Mafuuu's two options: make the typed-Enter lookup match a single chip too, or limit both texts to the two-key state and pin the single-chip result with a test.
  • CRF-162 merges Gon, Leorio, Mafuuu (all P4). Meruem's note and Razor's note say the case folding is correct for every Workspaces category (all backend lookups use lower()), so the finding is the stale comment, not the behavior. Leorio's alternative (gate listedOption on scopeToggle) is listed as the other option.
  • CRF-163 (Meruem): claims verified by reading. Two enterCategory dispatches (:836-839, :1214-1218) each call setTypedScopeWidened; leaveCategory and close do not reset it; isScopeWidened (:385) and appliedScopeChipFor (:415) mask a stale value with activeCategoryKey === category.key. No current bug. Kept at P4 as a structural finding: the PR adds the state, and CRF-102 (round 7) changed its contents, not where it lives, so this is not a re-raise.
  • CRF-164 to CRF-167 dropped (reasons in the rows). CRF-164 keep-argument: it verifies CRF-151. CRF-167 keep-argument: it names the assumption the comment should state; that is now part of CRF-162.
  • Out of scope: none new.

Post result: review 5324372078 (COMMENT), 4 inline comments; CRF-162 at useFilterCombobox.ts:1346 is not on the diff and was folded into the review body by post. No replies.

Reviewed against 3302857..cee3c66.

Round 13

Churn guard: PROCEED. CRF-159 to CRF-163 addressed in e3352f4 (CRF-162 had no thread because post folded it into the review body; the churn guard read the code), CRF-72 deferred to #29955, none silent. Head 7d90727; base ae1104f. Netero ran (post-panel round): 1 P4 (CRF-168). Law not spawned: effective additions 1717, growth 490 since round 5.

Fix round: e3352f4 only fixes round 12 findings (+166 -40 in 3 files), including moving typedScopeWidened into the reducer as CRF-163 asked; the other new commits merge the #29876 base (flyout rename to shownFlyoutCategory, read by Netero and Melody). Panel: personas that raised open findings the commit touches (bisky for CRF-160; razor, mafuuu for CRF-161; gon, leorio, mafuuu for CRF-162; meruem for CRF-163), ging-ts and ging-react for the changed .ts/.tsx files, nami for UI state, melody for the token pairing, kite (wildcard, random draw from the distilled set excluding mafuuu, meruem, razor). No contested findings.

Cross-check:

  • Prior fixes verified: CRF-159 (Netero M1), CRF-160 (Netero M4), CRF-161 (Netero M2, M3; Melody scratch tests), CRF-162 (Leorio, Netero by reading), CRF-163 (Netero M6, Meruem, Mafuuu). Ging TS, Ging React, Nami, Kite: no findings.
  • CRF-169 merges Meruem P4 and Melody P4, both reproduced with initialValue: "user:Alice" (row unchecked, click emits user:alice; typed Enter keeps user:Alice). Verified by reading :445-448: appliedScopeChipFor gates scopeChipHolding on length > 1. Meruem: removing the gate passes all 205 tests. The same exact-case row check in non-scope categories predates the PR (base FilterCombobox.tsx:1049, :1146), so that part is CRF-176, out of scope.
  • CRF-170 (Bisky): verified by reading. The listed branch runs only when scopeChipHolding(typed) returned nothing, and listedOption has the same lowercase value, so the lookup inside optionTokenFor returns nothing too. Interaction: CRF-169's gate removal leaves this redundancy in place.
  • CRF-171 merges four reviewers on the misplaced comment. Gon rated P2. Keep-argument for P2: the comment states the scope rule above two helpers that do not implement it, and four reviewers found it independently. Settled at P4: no runtime effect, a four-line move, and the other comment-accuracy findings in this PR (CRF-152, CRF-162) are P4.
  • CRF-172 merges Gon P4, Razor Nit, Leorio Nit; P4 by the tiebreak, and it is a contract ambiguity in the exported doc, not a style rule.
  • CRF-173 merges Gon Nit and Leorio Nit on the test name; distinct from CRF-168 (same test, weak assertion).
  • CRF-174 (Gon P2) dropped. Keep-argument: the reason for the case fold (backends compare lower(...)) is written nowhere in the code. Drop reason: the comment states what the function returns, the case rule, and the prefix exception, which is its contract; the case fold is pinned by two tests (Netero M1); CRF-120, CRF-121, and CRF-134 dropped the same pattern.
  • CRF-175 (Razor P4) dropped. Keep-argument: seven writes no test can observe are a third mechanism for one invariant (Netero M5 also found them unobservable). Drop reason: round 12 asked for the resets in leaveCategory and close (CRF-163), and they keep the field null outside a category, so a new read site without the activeCategoryKey guard stays correct. Same pattern as CRF-133 and CRF-153.
  • CRF-177 to CRF-180 dropped as cleared concerns.
  • Out of scope: CRF-176.

Post result: review 5324557962 (COMMENT), 6 inline comments, no replies.

Reviewed against cee3c66..7d90727.

Round 14

Churn guard: PROCEED. CRF-168 to CRF-173 addressed in 458fbb9, none silent. CRF-176 (out of scope) got no reply; out-of-scope items do not block. Head 8baceb7; base 9c006d2. Netero ran (post-panel round): 1 P3, 1 Nit (CRF-181, CRF-182). Law not spawned: effective additions 1725, growth 498 since round 5.

Fix round: 458fbb9 only fixes round 13 findings (+33 -33 in 3 files); the other new commits merge the #29876 base, whose onHighlightedValueChange("") call reaches the PR's early return in handleHighlightedValueChange (Netero, Melody). Panel: personas that raised open findings the commit touches (meruem, melody for CRF-169; bisky for CRF-170; gon, razor, mafuuu, leorio for CRF-171; gon, razor, leorio for CRF-172; gon, leorio for CRF-173), ging-ts and ging-react for the changed .ts/.tsx files, nami for UI state, chopper (wildcard, random draw from the distilled set excluding mafuuu, meruem, razor). No contested findings.

Cross-check:

  • Prior fixes verified: CRF-168 (Netero M1, Bisky), CRF-169 (Netero M2, Bisky, Mafuuu, Razor, Melody scratch tests), CRF-170 (Netero, Bisky by reading), CRF-172 and CRF-173 (Gon, grep). CRF-171's move holds, but it dropped a sentence (CRF-184). Meruem, Ging TS, Ging React, Chopper: no findings.
  • CRF-181 merges Netero P3 and Nami P3: both make typing sha fail to show the Owner flyout, through the same shownFlyoutKey expression. Netero's cause: the highlight condition, where typed text turns autoHighlight off and leaving the menu sets highlightedCategoryKey to null (reproduced after hovering Owner or Running). Nami's cause: flyoutCategoryKey ?? wins while another category's flyout is open and its row is hidden by the text (reproduced after hovering Template). One fix to the expression covers both, so one comment. Bisky's mutation confirms all four scope-flyout tests pass only because they render Owner alone. The highlight condition was already in the round 13 head (orchestrator read FilterCombobox.tsx:180-184 at 7d90727), so this is a first report, not a regression. Kept at P3: the PR description's sha path fails after ordinary pointer movement; entering Owner from its row still reaches the switch.
  • CRF-183 merges Gon P4 and Leorio P4 (both reproduced the flyout with no aria-selected row). Kept separate from CRF-181 (text, not behavior), and the comment tells the author to state whatever rule the CRF-181 fix settles.
  • CRF-184 merges Razor, Mafuuu, Leorio (all P4). Round 13's CRF-171 comment told the author to keep only the chip rule and the typed-key sentence, which dropped the typed-prefix sentence; the posted comment says so. Fourth instance of the suggested-text pattern (journal 2026-09-26).
  • CRF-186 (Gon P2) dropped. Keep-argument: the chip sentence reads the last two lines of the body aloud. Drop reason: three reviewers show the comment is wrong without the decision order, and the order includes the chip rule; a comment that states the order of the function's branches is its contract; CRF-174 precedent. Contradiction recorded: Gon wanted fewer words, the three wanted the typed-prefix sentence back; the comment gives the three's fix.
  • CRF-185 merges Gon Nit (line 2536) and Leorio Nit (reported 2507; the test name is at 2536 at head).
  • CRF-187 to CRF-191 dropped (reasons in the rows).
  • Out of scope: none new.

Post result: review 5324744454 (COMMENT), 5 inline comments, no replies.

Reviewed against 7d90727..8baceb7.

Round 15

Churn guard: PROCEED. CRF-181 to CRF-185 addressed in 8a1ce33 and 9a045a4, none silent. Head 9a045a4; base c0e978f. Netero ran (post-panel round): 1 P4 (CRF-192). Law ran: effective additions 1811, growth 584 since round 5 (above 500); verdict Don't split, Advisory.

Fix round: 8a1ce33 and 9a045a4 only fix round 14 findings; the other new commits merge the #29876 base (merge resolution in 6f17796 read by Netero and Melody). Panel: nami, gon, leorio, razor, mafuuu (the round 14 raisers of CRF-181, CRF-183, CRF-184, CRF-185), ging-ts and ging-react for the changed .ts/.tsx files, melody, bisky, meruem. No contested findings.

Cross-check:

  • Prior fixes verified: CRF-181 (Netero and Nami mutations, Melody scratch test, Mafuuu, Razor), CRF-182 (Netero), CRF-183 (Leorio, Mafuuu; its new null sentence is wrong, CRF-193), CRF-184 (Razor, Mafuuu, Leorio), CRF-185 (Leorio). Ging TS, Ging React, Nami, Leorio, Mafuuu, Melody: no findings.
  • CRF-192 merges Netero P4 (the condition at :287) and Gon Nit (the comment clause at :281 that describes it; Gon asks to delete it with the condition). One fix, so one comment at P4. Verified by reading :197-201 and :284-289: with shownFlyoutKey null, find on a null key returns undefined, the same as the condition.
  • CRF-193 merges Gon P4 (:183), Meruem P4 (:182), Razor P4 (:196). Orchestrator grep: setHighlightedCategoryKey is called only at :268 (undefined), :274, and :609 (onMouseLeave); close does not call it. Meruem and Razor reproduced the switch disappearing after the pointer leaves the menu with typed text; Razor quotes the PR description sentence "until another row is highlighted" (context.md line 21). Combined because one decision (should leaving the menu hide the flyout) fixes both the comment and the description or changes onMouseLeave. The comment names the invariant and both options instead of giving replacement text (journal 2026-09-26); Gon's replacement "another row or the pointer left the menu" was checked against the writers and is accurate only if the hiding stays.
  • CRF-194 (Bisky): the counter in onMouseLeave (0 after hover, 1 after user.type) is empirical and contradicts Razor's note that the tests cover hovering Running without leaving (CRF-199). Keep-argument for P4 over Nit: the row's name promises a case no test runs; the End test at :2914 pins the same mutation, but a test that claims coverage it does not have is a test-validity problem, not style.
  • CRF-195 (Bisky): mutation scopeMatchShown = !isCoarsePointer && scopeMatchKey !== null passes 222 tests; distinct from Nami's mutation (=== scopeMatchKey to false fails 2), which pins the show half, not the hide half.
  • CRF-196 (Gon Nit): kept; Leorio's note on test names covers the %s and Template names, not the :2914 name.
  • CRF-197 (Gon P2) dropped. Keep-argument: two of its sentences read the expression aloud, its last sentence is the third statement in the component that a flyout the text hides returns (:169-170, :265-266), and it does not say why undefined shows the flyout. Drop reason: CRF-183 asked the author to state the rule the CRF-181 fix settled, Leorio and Mafuuu verified the comment matches scopeMatchShown and shownFlyoutKey, each of the three "returns" sentences sits at a different mechanism (the reset, the "" branch, the non-narrowed branch), the reason for undefined is in the state comment at :183-184, and Gon's replacement ("Shows the scope match's flyout with no row highlighted") is false after onMouseLeave, where no row is highlighted and the flyout is hidden. CRF-174 and CRF-186 dropped the same pattern.
  • CRF-198 to CRF-200 dropped as cleared concerns.
  • The round summary had attributed CRF-193's Meruem half to a persona not on this panel; corrected by re-reading meruem.md before writing the inventory.
  • Out of scope: none new.

Post result: review 5324967231 (COMMENT), 5 inline comments, no replies.

Reviewed against 8baceb7..9a045a4.

Round 16

Churn guard: PROCEED. CRF-192 to CRF-196 addressed in 710a26d, CRF-72 deferred to #29955, none silent. The churn guard called every other new commit a base commit; 02631cd (typed-Enter wording in types.ts and useFilterCombobox.ts) is PR-only (orchestrator: git merge-base --is-ancestor 02631cd40f 6972d68f7d fails) and was given to the panel in context.md. Head 2fc73a2; base 6972d68. Netero ran (post-panel round): 1 P3, 1 P4, 1 Nit (CRF-201 to CRF-203). Law not spawned: effective additions 1846, growth 35 since the round 15 analysis.

Inventory validation: CRF-197 reopened, because 710a26d makes onMouseLeave set undefined, which removed two of its drop reasons (the comment matching the code, Gon's replacement being false). No other non-posted disposition depends on the changed code; the base's shownOptions change and the first-match highlight inside a category were given to the panel as facts.

Fix round: 710a26d fixes round 15 findings and 02631cd rewords two comments; merges d079c80 and 2fc73a2 bring the #29876 base (d079c80 resolved a conflict in FilterCombobox.tsx). Not a restructure: no files added or moved, no PR signature change, same goal, growth 35. Panel: raisers of open findings the commits touch (gon for CRF-192, CRF-193, CRF-196, reopened CRF-197; meruem, razor for CRF-193; bisky for CRF-194, CRF-195), leorio (once-per-PR floor rule: 02631cd touches a doc comment), ging-ts and ging-react for the changed .ts/.tsx files, nami (frontend UI state), melody (writer and reader pairing of highlightedCategoryKey), knov (wildcard, random draw from mafuuu, chopper, kite, zoro, knov). No contested findings.

Cross-check:

  • Prior fixes verified: CRF-192 (Netero, Razor), CRF-193 (Netero, Meruem, Razor, Melody, Nami mutations and scratch tests; Leorio by reading), CRF-194 and CRF-195 (Bisky probes and mutations, Netero), CRF-196 (Netero, Leorio). Ging TS, Ging React, Knov, Bisky, Nami, Melody: no findings.
  • CRF-201 (Netero P3) verified by reading FilterCombobox.tsx:291-299 and the remerge diff of d079c80. Keep-argument for P3: the merge put the live-region rule above the scope toggle, and a merge resolution is a hunk no one reviews as a change. Settled at P4: a two-line move with no runtime effect, the same class as CRF-171 (misplaced comment, P4).
  • CRF-202 (Netero P4): mutation evidence (disabling listedOption passes 197 of 197 at head, fails both rows at round 15). The branch is reachable: Gon, Meruem, and Leorio each cleared the highlight inside Owner with hover and unhover. The comment gives only the test fix, because the PR description documents the listed-owner rule the branch implements.
  • CRF-203 (Netero Nit): verified by reading types.ts:66. Netero's section evidence also measured line 70 at 82 columns; the comment names it.
  • CRF-204 merges Razor P4 (:188) and Gon P2 (:187, filed as CRF-197 re-raised). Split from CRF-197 because the false condition is new with 710a26d (CRF-193's comment asked to update lines 188-191; the author did not). Keep-argument for P2: the comment contradicts a test added in the same commit, and round 15 asked for this update. Settled at P4: the test pins the behavior, so leaving the comment costs little (P4 definition), and every comment-accuracy finding in this PR is P4 (CRF-152, CRF-162, CRF-171, CRF-183, CRF-184, CRF-193). The comment gives Razor's phrase replacement, checked against the undefined writers (the "" pick, onMouseLeave, initial state) and the :181-183 definition.
  • CRF-197 remaining parts dropped again. Keep-argument: "never shows on coarse pointers" reads !isCoarsePointer aloud and the last sentence is the third copy of the text-hides-flyout rule. Drop reason: each copy sits at a different mechanism; the comment states the rule across narrowing, highlight, and pointer type; Gon's new replacement omits the narrowed-text and coarse-pointer conditions; CRF-174 and CRF-186 precedent.
  • CRF-205 merges Gon P2 and Leorio P4 (both :279). Our round 15 CRF-192 comment said to delete the condition and one clause; that left two sentences that describe no code under them. Fifth instance of the suggested-text pattern (journal 2026-09-26). Keep-argument for P2: two of three sentences describe code elsewhere, and the click rule's actual site (drillIn at :604-607) carries no comment. Settled at P4, CRF-171 precedent. The Enter rule's site (useFilterCombobox.ts:1582-1591) has no comment, so the comment says to move the sentences, not delete them.
  • CRF-206 (Gon P2) dropped. Keep-argument: it lists two setters a reader can see, omits the initial state, and "another row" can be read as another category row, whose value is its key. Drop reason: "as when" marks examples, so omitting the initial state is not false; Leorio, Meruem, and Melody verified it matches every writer; the first words "Highlighted category row" set the contrast for "another row"; round 15 asked for this update (CRF-193); CRF-174, CRF-186 precedent. Contradiction recorded: Gon versus Leorio and Meruem notes.
  • CRF-207 merges Gon, Meruem, Leorio (all P4), each reproduced with hover and unhover inside Owner before Enter. Razor's and Melody's notes confirm the types.ts and scopeChipHolding wording matches !hasHighlightedOption, so the description is the outlier. Line from Gon (useFilterCombobox.ts:1496, the hasHighlightedOption check). The replacement "no option highlighted" was checked against the !activeOptionsLoading guard: while loading, Enter commits nothing, so the chip is still kept.
  • CRF-208 (Gon Nit) kept as a wording Nit; the comment does not give replacement text.
  • CRF-209 merges Leorio Nit (:3047) and Gon Nit (:3020); both name both tests. Verified by reading FilterCombobox.test.tsx:3020-3065: both type sha, and the :3020 test needs sha to match sharon.
  • CRF-210 to CRF-220: Gon's pattern note dropped; Meruem's out-of-scope item is CRF-211 (typed Enter never toggles, from the base, and the first-match highlight is base behavior); the rest are cleared-concern notes.

Post result: review 5325153009 (COMMENT), 8 inline comments, no replies.

Reviewed against 9a045a4..2fc73a2.

Round 17

Churn guard: PROCEED. CRF-201 to CRF-205 and CRF-207 to CRF-209 addressed (43743c8; CRF-207 by a PR description edit), CRF-72 deferred to #29955 (the churn guard's gh issue view failed with HTTP 401, so its current state is unverified this round), none silent. Head 43743c8; base ca73fb7. Netero ran (post-panel round): 1 Nit (CRF-221). Law not spawned: effective additions 1850, growth 39 since the round 15 analysis.

Inventory validation: no non-posted disposition depends on code 43743c8 or the c69eaf1 merge resolution changed.

Fix round: 43743c8 only answers round 16 findings (comments moved and reworded, one test changed); merges ec6944f and c69eaf1 bring the #29876 base (c69eaf1 resolved a conflict in CategoryOptionsList emptyMessage). Not a restructure: no files added or moved, no PR signature change, same goal, growth 4 since round 16. Panel: raisers of open findings the commit touches (gon for CRF-204, CRF-205, CRF-207 to CRF-209; leorio for CRF-205, CRF-207, CRF-209; razor for CRF-204; meruem for CRF-207), bisky (the CRF-202 test change), ging-ts and ging-react for the changed .ts/.tsx files, nami (frontend UI state), zoro (wildcard, random draw from mafuuu, chopper, kite, zoro, knov). No contested findings.

Cross-check:

  • Prior fixes verified: CRF-201, CRF-203, CRF-205, CRF-208, CRF-209 (Netero, Gon, Leorio, Meruem, Razor by reading), CRF-202 (Netero, Razor, Bisky, Nami mutations), CRF-204 (Razor; its wrap is CRF-221), CRF-207 (Leorio, churn guard read the description). Leorio, Razor, Meruem, Bisky, Ging TS, Ging React, Nami, Zoro: no findings.
  • CRF-221 (Netero Nit): verified in the 43743c8 diff (line 189 holds two sentences past column 80). Third wrap Nit after a text fix in this PR (CRF-182, CRF-203).
  • CRF-222 (Gon P3) dropped. Keep-argument: the base comment's reason ("instead of opening a flyout") was lost through the PR's edits, and the reason a click must drill in while text narrows (shownFlyoutKey ignores flyoutCategoryKey) is written nowhere. Drop reason: Gon's severity rested on "removing || categoriesNarrowedByText passes 200 of 200 tests"; the orchestrator ran that mutation at 43743c8 and 2 tests fail ("opens a category narrowed by typed text when clicked", "drops the typed text when a row listed by the scope phrase is clicked"), so the empirical result wins; the comment is accurate (Leorio, Meruem, Razor) and sits where round 16 asked.
  • CRF-223 (Gon P2) dropped. Keep-argument: the comment reads the condition aloud, and the reason the text stays a search (text such as sha may be a workspace name search, CRF-11) is not in the code. Drop reason: round 16 asked for the sentence at this branch (CRF-205); it states the rule, which Leorio and Meruem verified; the test "keeps free text when Enter targets a row listed by the scope phrase" pins the behavior; CRF-174 precedent. Gon's second point ("scope phrase" for searchPhrase) is CRF-225's reason.
  • CRF-224 (Gon P2) dropped. Keep-argument: the sentence reads !isMobile aloud and ignores activeCategoryKey === null. Drop reason: round 16's comment told the author to keep only this sentence (CRF-205); reversing that request one round later is churn; same pattern as CRF-133 and CRF-153.
  • CRF-225 (Gon Nit) dropped. Keep-argument: the field is searchPhrase, and "scope" also names the switch. Drop reason: the component's identifiers call typed text that matches searchPhrase a scope match (findScopeMatch, scopeMatchKey, scopeMatchedCategory, SCOPE_MATCH_MIN_QUERY_LENGTH), so "scope phrase" follows that vocabulary.
  • CRF-226 to CRF-228 dropped as cleared-concern notes.
  • Out of scope: none new.

Post result: review 5325230124 (COMMENT), 1 inline comment, no replies.

Reviewed against 2fc73a2..43743c8.

Round 18

Churn guard: PROCEED. CRF-221 addressed in 2d06c84; CRF-72 deferred to #29955 (gh issue view failed with HTTP 401 again, so its current state is unverified), none silent. Head 2d06c84; base b897508. Netero ran (post-panel round): no findings. Law not spawned: effective additions 1851, growth 40 since the round 15 analysis.

Inventory validation: no non-posted disposition depends on the rewrapped comment or the merged base lines (Netero checked the same).

Fix round: 2d06c84 only rewraps the scopeMatchShown comment (text unchanged); merge c985108 brings the #29876 base with no conflict (base changes: fakeTimers: "manual" test helper, typeaheadOptionsByKey iterates categories). Not a restructure. Panel: gon (a comment changed), ging-ts and ging-react for the changed .tsx file, nami (frontend), knov (wildcard, random draw from the distilled set; none of it was selected). CRF-221's raiser is Netero, which ran. No contested findings.

Cross-check:

  • CRF-221 fix verified (Netero, Gon, Nami). Gon, Ging TS, Ging React, Nami, Knov, Netero: no findings. Gon checked that the rewrap is not new evidence against the CRF-197 drop.
  • Netero and Nami traced the base loop change: the only PR consumer of typeaheadOptionsByKey (collectValueSuggestions) iterates menuCategories, and the PR's categories wrapper keeps every key.
  • CRF-229 and CRF-230 dropped as cleared-concern notes.
  • No open in-scope findings remain; CRF-72 is deferred with a ticket. Review event APPROVE (the CLI downgrades it to COMMENT).
  • Out of scope: none new.

Post result: review 5325309053, posted in APPROVE mode with no inline comments (the CLI output does not show whether GitHub recorded APPROVE or the COMMENT downgrade). Dismissing the earlier REQUEST_CHANGES review failed: GET https://api.github.com/user: 403 Resource not accessible by integration.

Reviewed against 43743c8..2d06c84.

Round 19

Churn guard: PROCEED. No PR-only commits since round 18; CRF-72 deferred to #29955 (gh issue view failed with HTTP 401 again, so its state is unverified). Head c43b205; base 110ff08. Netero ran (post-panel round): no findings. Law not spawned: effective additions 1851, growth 40 since the round 15 analysis.

Inventory validation: the PR's own +/- lines match round 18 (orchestrator and Netero diffed them), so no non-posted disposition depends on changed PR code.

Fix round with no PR commits: merges 9141926 and c43b205 bring the #29876 base; 9141926 resolved a conflict in FlyoutCategoryPanel (base filteredOptions.map, PR optionTokenFor(option)). Not a restructure. Panel (at least 3 required for a fix round): nami (frontend UI state), ging-ts and ging-react for the .ts/.tsx files the merges changed, melody (the base's shownOptions output against the PR's readers), mafuuu (wildcard, random draw from the distilled set). No open findings whose code changed; no contested findings.

Cross-check:

  • Netero, Nami, Ging TS, Ging React, Melody, Mafuuu: no findings. Netero and Melody traced the merge resolution and the three base changes the PR reads (shownOptions with failed, activeOptions, typeaheadOptionsByKey); each gives the PR code the same values as at round 18. Unit tests 233 of 233 and story tests 55 of 55 (Netero).
  • CRF-231 dropped as a cleared-concern note.
  • No open in-scope findings remain; CRF-72 is deferred with a ticket. Review event APPROVE.
  • Out of scope: none new.

Post result: review 5325600359, posted in APPROVE mode with no inline comments. Dismissing the earlier REQUEST_CHANGES review failed again with the 403 integration error.

Reviewed against 2d06c84..c43b205.

Round 20

Churn guard: PROCEED. No open findings besides CRF-72 (deferred to #29955; gh issue view failed with HTTP 401, state unverified). Head 9237e4d; base 219e83e. One PR-only commit, 53e9abf, changes Owner option behavior without a finding asking for it, plus a new PR description bullet. Netero ran (post-panel round): 1 P4, 1 Nit (CRF-232, CRF-233), 1 out-of-scope item (CRF-236). Law not spawned: effective additions 1859, growth 48 since the round 15 analysis.

Inventory validation: no non-posted disposition depends on the self-row change. CRF-137 and CRF-180 (permission-set tests) use an empty query, for which the self row is unchanged; the vault note on typing one's own username still holds, because the label contains the username.

Fix round: 53e9abf is a one-line production change plus a test. Not a restructure (no files added or moved, no signature change, same goal, growth 8). Panel: ging-ts and ging-react for the changed .ts/.tsx files, nami (frontend options), bisky (new test, behavior change), leorio (new PR description bullet and commit text), melody (the loader's output against every Owner option consumer), razor (the description's "like every other owner" claim; two implementations of the match rule), zoro (wildcard, random draw from mafuuu, meruem, chopper, kite, zoro, knov). No contested findings.

Cross-check:

  • CRF-232 (Netero P4): scratch test evidence (getUserFilterOptions("smith", alice, users [alice, bob]) returns only bob) and users.sql:450-453 cited. Bisky confirms the new test's fake client ignores the query, so it cannot model this. Melody's and Nami's notes (the self row matches the local filter rule, so it does not appear and then vanish) are about the current rule, not a contradiction; with the fix, a name or email match appears when results arrive, as other users matched by email do. The comment does not claim a typed-Enter result for this case, because the first listed match would be highlighted, which no one reproduced.
  • CRF-233 (Netero Nit): awk measured line 77 at 93 columns; the line is PR-added (not in the base story file).
  • CRF-234 (Leorio P4): verified: the base defined a User category and OrdinaryUserKeepsUserChip; the PR renamed that story to OrdinaryUserGetsOwnerChip and left OrdinaryUserSeesOwnerAndUser and its comment unchanged. In scope: the PR's removal of User makes the name and comment wrong.
  • CRF-235 merges Leorio P4 and Zoro Nit (both line 99). P4 by the tiebreak: Leorio names the undocumented match rule and the repeat of WorkspacesFilter.tsx:53; the stale caller list is a contract statement, not a style rule. The comment does not give replacement text and tells the author to write the rule after settling the name and email case, in its own sentence (journal 2026-09-26 on follow-ups in either/or fixes).
  • CRF-236 (Netero out of scope): getSelfUserFilterOptions and its hand-written match exist in the base (219e83e0e6).
  • CRF-237 to CRF-240 dropped as cleared-concern notes.

Post result: review 5325651430 (COMMENT), 3 inline comments, no replies. CRF-234 at WorkspacesFilter.stories.tsx:107 is not on the diff; with side omitted, post --dry-run resolved it to LEFT line 107, a different line of the base file, so the orchestrator set side: RIGHT (Leorio's line is the head file's comment, verified by reading), and post folded it into the review body.

Reviewed against c43b205..9237e4d.

Round 21

Churn guard: PROCEED. CRF-232 to CRF-235 addressed in PR-only commit f8773ec (3 files, +26/-8); no silent items. CRF-72 deferred to #29955 (gh issue view failed with HTTP 401, state unverified). Head f8773ec; base 219e83e (no new base commit since round 20). Netero ran (post-panel round): no findings; verified CRF-232 to CRF-235 fixed; the new test fails with the fix reverted; tsc --noEmit exit 0, Biome clean on 14 files, vitest 4 files passed (21 + 10 + 3 + 201 tests). Law not spawned: effective additions 1881, growth 70 since the round 15 analysis.

Inventory validation: f8773ec changes only getUserFilterOptions, its test, and one story; no non-posted disposition depends on them. CRF-236 (out of scope) still holds: the fallback still calls getSelfUserFilterOptions with its hand-written match.

Fix round, not a restructure. Panel: gon (comment and name changes), leorio (CRF-234 and CRF-235 raiser, story rename), zoro (CRF-235 Nit raiser), bisky (new test), ging-ts and ging-react (changed .ts/.tsx files), nami (frontend Owner options), mafuuu (wildcard, random draw). No contested findings.

Cross-check:

  • CRF-241 merges Gon Nit and Leorio Nit on categoryOptions.test.ts:37. Verified by reading: the test at :44 lists me for smith, which is not in alice (you) or me, while :37 passes users: [] and so covers only the getSelfUserFilterOptions fallback. In scope: PR-added test whose name the round's commit made false. The comment's example name was checked against both assertions of the test (ali listed, zzz not listed).
  • CRF-242 (Gon Nit) dropped. Keep-argument: self names an array of zero or one options next to an array named options, so it reads as one option, and a leftover reference after a rename would resolve to the DOM global self. Drop reason: orchestrator scratch check, [...self, ...options] compiled against the DOM lib fails with TS2488 (Window & typeof globalThis has no iterator), so the only reference cannot resolve to the global without a type error; what remains is a naming preference with no project rule or lint check, and [selfUserOption(me)] in the same statement shows the array.
  • CRF-243 (Ging React P4) dropped. Keep-argument: the file focuses the same input two ways, and the DOM lookup does nothing, with no error, if the switch renders outside the cmdk root, while actions.focusInput reaches the input through a ref. Drop reason: in scope (added in PR commit 3e0145a, unchanged since), but the stated consequence does not hold. (1) The panel renders in-flow because disablePortal is required for cmdk to DOM-query its list (primitives.tsx:113), so a portal would break cmdk with or without this lookup. (2) The break is not silent: orchestrator mutation replacing [cmdk-root] with an unmatched selector fails "returns from the scope switch to the owner options with the arrow keys" (FilterCombobox.test.tsx:3203, assertion at :3219); the worktree was restored. (3) SkillsTriggerMenu.tsx:125 uses the same cmdk attribute lookup in production, though there cmdk exposes no ref, so this precedent is weaker. The fix threads a new prop through CategoryOptionsList, OptionsPanel, and FlyoutScopeToggle for no behavior change, and removes no documented tradeoff.
  • CRF-244 to CRF-246 dropped as cleared-concern notes.
  • Zoro, Ging TS, Mafuuu, Netero: no findings.
  • Out of scope: none new.

Post result: review 5325715032 (COMMENT), 1 inline comment (CRF-241 at categoryOptions.test.ts:37, RIGHT, on the diff), no replies. post --dry-run passed validation before posting.

Reviewed against 9237e4d..f8773ec.

Round 22

Setup restart: the first deep-review setup captured head f8773ec, but git ls-remote then showed the PR head at 1142d5e (the author's CRF-241 reply names it; the push landed after setup's head check). With no reviewers selected and nothing posted, the orchestrator moved that setup to stale-round-22-setup-at-f8773ecb3e/ (kept, not deleted), pointed current back at round-21, ran deep-review cleanup for the new orchestrator worktree, and reran deep-review setup, which created round 22 at 1142d5e.

Churn guard: PROCEED. CRF-241 addressed in PR-only commit 1142d5e (one line, the test name); no silent items. CRF-72 deferred to #29955 (gh issue view failed with HTTP 401, state unverified). Head 1142d5e; base 219e83e (unchanged). Netero ran (post-panel round): no findings; verified CRF-241 fixed; vitest 10 of 10 on categoryOptions.test.ts, Biome clean. Law not spawned: effective additions 1881, growth 70 since the round 15 analysis.

Inventory validation: 1142d5e changes only a test name; no non-posted disposition depends on it.

Fix round, not a restructure (one-line test rename, no files added or moved, no signature change, same goal, growth 0). Panel: gon and leorio (raised CRF-241, whose line the commit changes), ging-ts (.ts file changed), nami (file under site/src), bisky (test file changed), zoro (wildcard, shuf draw from mafuuu, meruem, chopper, kite, zoro, knov, razor). No contested findings.

Cross-check:

  • All six panel reviewers and Netero: no findings. Gon, Leorio, Nami, Bisky, Zoro, and Netero each read the renamed test against categoryOptions.tsx:128-130 and ran it (10 of 10).
  • CRF-247 merges the four matching notes as a cleared concern.
  • No open in-scope findings remain; CRF-72 is deferred with a ticket. Review event APPROVE.
  • Out of scope: none new.
About deep-review

CRF = Coder Review Finding (P0-P4, Nit, Note)

Reviewer Focus
Bisky tests
Chopper ops/errors
Churn-guard change verification
Ging language modernization
Gon naming
Hisoka edge cases
Killua perf
Kite change integrity
Knov contracts
Knuckle SQL
Komugi flake/determinism
Kurapika security
Law decomposition
Leorio docs
Luffy product
Mafu-san process
Mafuuu contracts
Melody dispatch/pairing
Meruem structural
Nami frontend
Netero mechanical checks
Pariston premise testing
Pen-botter product gaps
Razor verification
Robin duplication
Ryosuke Go arch
Takumi concurrency
Zoro shape

🤖 Managed by Coder Agents.

@tracyjohnsonux
tracyjohnsonux marked this pull request as ready for review September 24, 2026 23:34

@coder-agents-review coder-agents-review Bot 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.

This PR replaces the User category on /workspaces with an include-shared switch on the Owner category, adds a scope pill and a shared typeahead, and removes the Shared attribute. Findings: 4 P2, 13 P3, 4 P4, 6 Nit.

Out of scope (needs a ticket or explicit acceptance by a human):

  • site/src/components/Filter/FilterCombobox/useFilterCombobox.ts:957: with the menu open, Enter on a category row matched by typed text (for example sta for Status) opens that category and clears the free-text search, so a search that starts with a category name cannot be submitted with Enter.
  • site/src/components/Filter/FilterCombobox/filterQuery.ts:29: categoryPreview has no production caller and builds tokens from category.key, so it would not match user: chips if it were used again.
  • coderd/searchquery/search.go:283: the workspace query language has no OR, so "owned by or shared with" needs the fused user: key and a frontend key rewrite instead of a composable filter.
  • site/src/pages/WorkspacesPage/filter/WorkspacesFilter.tsx:53: "Users who cannot list others only see themselves." repeats the TODO above it; the repetition predates this PR.

Notes:

  • Existing shared:true URLs now render as free text instead of a chip. The backend still applies the filter.
  • scopeToggle is a generic FilterCombobox option with one consumer, and about 360 lines of production code in FilterCombobox.tsx and useFilterCombobox.ts serve it.

site/src/components/Filter/FilterCombobox/types.ts:14

P3 [CRF-4] The token doc comment still lists shared:true as an Attributes token, and no category commits it after this PR. (Netero)

Replace the example with the two remaining tokens, outdated:true and dormant:true.

🤖

site/src/pages/WorkspacesPage/filter/categoryOptions.tsx:99

Nit [CRF-28] Two comments in categoryOptions.tsx still describe separate User and Owner categories. (Melody, Luffy)

Lines 99 and 100 ("so the User and Owner categories stay available") and line 118 ("Shared by the User and Owner categories") now describe one Owner category with the owner and user keys. Update both to match the comment at WorkspacesFilter.tsx:51.

🤖

🤖 This review was automatically generated with Coder Agents.

…o fix/filter-combobox-mouse-interactions

# Conflicts:
#	site/src/components/Filter/FilterCombobox/FilterCombobox.tsx
#	site/src/components/Filter/FilterCombobox/queries.ts
#	site/src/components/Filter/FilterCombobox/types.ts
#	site/src/components/Filter/FilterCombobox/useFilterCombobox.ts
@tracyjohnsonux
tracyjohnsonux changed the base branch from fix/filter-combobox-mouse-interactions to feat/filter-combobox-clear-all September 25, 2026 02:03
…r-combobox-hide-single-option

# Conflicts:
#	site/src/components/Filter/FilterCombobox/useFilterCombobox.ts
…ox-owner-shared-scope

# Conflicts:
#	site/src/components/Filter/FilterCombobox/FilterCombobox.tsx
#	site/src/components/Filter/FilterCombobox/filterQuery.ts
#	site/src/components/Filter/FilterCombobox/useFilterCombobox.ts
…r-combobox-hide-single-option

# Conflicts:
#	site/src/components/Filter/FilterCombobox/types.ts
#	site/src/components/Filter/FilterCombobox/useFilterCombobox.ts
@tracyjohnsonux

Copy link
Copy Markdown
Collaborator Author

CRF-234 fixed in f8773ec: renamed the story to OrdinaryUserSeesOwner and dropped User from its comment.

Generated by Coder Agents

@tracyjohnsonux

Copy link
Copy Markdown
Collaborator Author

/coder-agents-review

@coder-agents-review coder-agents-review Bot 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.

This PR merges the User and Owner workspace filters into one Owner category with a shared-scope switch. This round: 1 Nit.

f8773ec fixes the four round 20 findings: the {username} (you) row stays listed when the users API returns the current user, the story comment is wrapped, the story is renamed to OrdinaryUserSeesOwner, and the getSelfUserFilterOptions comment states its match rule.

🤖 This review was automatically generated with Coder Agents.

@tracyjohnsonux

Copy link
Copy Markdown
Collaborator Author

/coder-agents-review

@coder-agents-review coder-agents-review Bot 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.

This PR merges the User and Owner workspace filters into one Owner category with a shared-scope switch. This round: no new findings.

1142d5e fixes the round 21 finding: the test name now states that it covers the label match used when the users API does not return the current user. No in-scope finding from earlier rounds remains open; the user guide update is tracked in #29955.

🤖 This review was automatically generated with Coder Agents.

…box-remove-workspace-previews

# Conflicts:
#	site/src/components/Filter/FilterCombobox/FilterCombobox.tsx
#	site/src/components/Filter/FilterCombobox/types.ts
#	site/src/pages/WorkspacesPage/filter/WorkspacesFilter.tsx
…workspace-previews' into fix/filter-combobox-mouse-interactions

# Conflicts:
#	site/src/components/Filter/FilterCombobox/FilterCombobox.tsx
#	site/src/components/Filter/FilterCombobox/primitives.tsx
#	site/src/components/Filter/FilterCombobox/useFilterCombobox.ts
…-option' into feat/filter-combobox-clear-all

# Conflicts:
#	site/src/components/Filter/FilterCombobox/primitives.tsx
…into feat/filter-combobox-owner-shared-scope

# Conflicts:
#	site/src/components/Filter/FilterCombobox/primitives.tsx
#	site/src/components/Filter/FilterCombobox/types.ts
#	site/src/pages/WorkspacesPage/filter/categoryOptions.tsx
…ouse-interactions

# Conflicts:
#	site/src/components/Filter/FilterCombobox/FilterCombobox.stories.tsx
#	site/src/components/Filter/FilterCombobox/FilterCombobox.tsx
#	site/src/components/Filter/FilterCombobox/queries.ts
#	site/src/components/Filter/FilterCombobox/types.ts
#	site/src/components/Filter/FilterCombobox/useFilterCombobox.ts
tracyjohnsonux added a commit that referenced this pull request Sep 28, 2026
…nline options (#29426)

## Summary

Improves mouse and touch interactions for the workspace FilterCombobox
while preserving its keyboard navigation.

The filter control (icon only) now opens a compact main menu with inline
status and workspace-state options, pointer flyouts for Owner, User,
Template, and Organizations, and a single-panel mobile flow. Applied
chips always render lowercase. Selected workspace attributes remain
single-select to match the existing workspace filtering UI.

Stacked on #29874 (removes the workspace result previews). Follows
#29111, which has merged. Split out and stacked on this PR: #29875 (hide
single-option categories), #29876 (Clear all), and #29872 (Owner shared
workspaces toggle).

<details>
<summary>Implementation notes</summary>

- Keeps Up/Down, ArrowRight, ArrowLeft, Backspace, Escape, and Tab
keyboard behavior. Free-typed text no longer highlights the first row:
Enter applies it as a search and closes the menu, while arrow keys,
Home/End, cmdk's Ctrl+N/J/P/K, or the pointer pick a filter row. Tab
still completes the top row, and a typed `key:` prefix, the category
view, and the all-filters list keep the first match highlighted.
- Enter activates any focused button in the combobox (Filters, flyout
options, Retry, chip remove) through one `FilterComboboxRoot` handler.
- The menu's height is bounded only by the viewport, so it scrolls only
when it cannot fit.
- Uses right-opening pointer flyouts with hover forgiveness and
searchable, scrollable long lists. The flyout follows the cmdk
highlight, so keyboard navigation closes or switches it, and it resets
whenever the menu opens or closes.
- Inline categories (Status, Workspace attributes) never enter category
mode; a typed `status:` prefix narrows the inline group in place and is
withheld from the workspace search query. A failed inline load shows its
error and Retry under the category heading, since inline categories have
no flyout.
- Browsing and category mode share one main-panel renderer.
- Uses a persistent full-width mobile panel with touch scrolling. While
open on mobile the field pins below the navbar with CSS (`fixed`), with
no JS scroll positioning. Verified in Chrome device emulation, not yet
on a physical device with a software keyboard.
- Keeps free-text workspace searches separate from matching
filter-option text, including text typed before a `status:` prefix.
Typed text that could still match a filter (a category name, a loaded
option, or an option the category's search returns) is held back from
the workspace search, so results do not empty out mid-word. It is
applied when no lookup finds a match (a failed lookup counts as no
match, and the wait stops after 1s), on Enter, or when the menu is
dismissed or switched to all filters. The menu shows no separate cue for
held-back text; the author (@tracyjohnsonux) decided this, and also
rejected a `Search workspaces for "..."` row. That is why Enter-search
is built by suppressing cmdk's automatic first-row highlight
(`userNavigatingRef` and the `""`/`" "` reset in `FilterComboboxRoot`)
rather than by rendering a search row. Chip tokens in the typed text
become chips when it is applied.
- While typed text is debounced, loaded options are filtered locally so
matching rows stay visible until the loader's results arrive.
- Enter or Tab on an applied option that typed text located keeps it and
clears the text. Clicking a checked row removes it.
- Enter commits a typed inline value that is not a suggestion, such as
`status:starting`.
- Flyout searches go through the category loader, so owners beyond the
first page are found. The search field stays mounted while results load,
shrink, or are empty.
- Clicking a category row narrowed by typed text enters the category,
and focus returns to the input after picking a flyout option.
- Keeps selected options visible with Coder-standard checkmarks and
styling.
- `FilterCategory.chipLabelOnly` marks categories whose chips show only
the option label (Workspace attributes). Inline options load up front so
those chips are labelled before the menu is first opened.
- Pointer flyouts and category lists share one `OptionsPanel` shell
(search header, list, empty state).

</details>

Generated by Coder Agents.

---------

Co-authored-by: Jake Howell <jacob@coder.com>
Co-authored-by: Jake Howell <jake@hwll.me>
…hide-single-option

# Conflicts:
#	site/src/components/Filter/FilterCombobox/FilterCombobox.stories.tsx
#	site/src/components/Filter/FilterCombobox/FilterCombobox.test.tsx
#	site/src/components/Filter/FilterCombobox/FilterCombobox.tsx
#	site/src/components/Filter/FilterCombobox/primitives.tsx
#	site/src/components/Filter/FilterCombobox/types.ts
#	site/src/components/Filter/FilterCombobox/useFilterCombobox.ts
…clear-all

# Conflicts:
#	site/src/components/Filter/FilterCombobox/FilterCombobox.stories.tsx
#	site/src/components/Filter/FilterCombobox/FilterCombobox.tsx
#	site/src/components/Filter/FilterCombobox/useFilterCombobox.ts
Base automatically changed from feat/filter-combobox-clear-all to main September 28, 2026 23:51
@tracyjohnsonux
tracyjohnsonux merged commit 749fb9e into main Sep 29, 2026
31 checks passed
@tracyjohnsonux
tracyjohnsonux deleted the feat/filter-combobox-owner-shared-scope branch September 29, 2026 17:45
@github-actions github-actions Bot locked and limited conversation to collaborators Sep 29, 2026
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants