Repository navigation
fix(plugin-chatbot): fence chatbot-floating's raw props spread - #8077
Merged
Merged
Conversation
chatbot-floating ended its <FloatingChatbot> element with a raw
{...props} spread, LAST, where its two sibling registrations (chatbot,
chatbot-enhanced) spread {...toDomProps(props)} FIRST. Per the card's
ruling (fence, not declare), moving it to the head and filtering it
through toDomProps closes all three consequences at once: a sent
message now renders on a floating chatbot (the authored `messages`
seed no longer overrides the live runtime messages), processVisibility
/ surface / showAvatars go dark on chatbot-floating nodes (matching
what ChatbotFloatingSchema has always declared), and displayMode /
systemPrompt / model stop leaking as DOM attributes on the panel root.
The three tripwire pins in renderer.authoring-faces-7655.test.tsx
(section 4) flip from lit to dark, deliberately, as planned when they
were written. New coverage owns the two consequences that were not
pinned anywhere before this fix: renderer.floating-spread-fence-7708.test.tsx.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01YBWFb5YgMU5dw8p2VKj16S
Contributor
✅ Console Performance Budget
The eager closure is every chunk the entry reaches through static imports — what the browser fetches and parses before the app renders. The entry chunk on its own is a small fraction of it. 📦 Bundle Size Report
Size Limits
|
Collaborator
Author
Landing — in-seat contract review PASSED; standing down on one red that is not this PR'sBy the dispatching seat (
|
job step 7 Start ObjectStack backend (published packages) |
job | |
|---|---|---|
main 09-05, run 33950494089 job 101264218858 |
✅ success in 76s | ✅ success — the lit control |
main 09-06, run 34017174769 job 101442890465 |
❌ failure after ~6 min | ❌ failure, steps 8–14 all skipped |
conclusion: success at the RUN level while its job fails. ⛔ Never read a workflow run's conclusion as evidence in this repo.
No fix exists to port — the break is undiagnosed and carded as objectui#8084, with the published-package hypothesis named as the first thing to test. ⛔ No re-run is being spent here: this is a reproducible base-red, not a flake, and a re-run would only confirm what two main runs already establish.
⇒ Per the standing rule, this is one comment naming the failing check and why it is not this PR's, and then the PR proceeds. Stripping needs:contract-review, flipping ready, enqueuing.
Generated by Claude Code
Collaborator
Author
|
Collaborator
Author
Follow-up to the correction:
|
os-justin
deleted the
claude/issue-7708-chatbot-floating-spread-fence
branch
September 6, 2026 15:26
This was referenced Sep 6, 2026
os-justin
pushed a commit
that referenced
this pull request
Sep 6, 2026
loading, showAvatars, userAvatar, assistantAvatar, markdown and height were declared on ChatbotSchema, mirrored on its Zod twin, and read by no plugin-chatbot registration: a schema.KEY census per ComponentRegistry.register body returns 0/0/0 for all six, with placeholder 1/1/1, messages 1/1/1, userAvatarUrl 1/1/1, maxHeight 1/1/0, floatingConfig 0/0/1 and processVisibility 0/1/0 lit on the same instrument. Each becomes `?: never` on complex.ts plus retirementTombstone() on complex.zod.ts — both halves, the convention #6972 / #6355 / #7779 already carry. Deleting them was the wrong route: all six have a Zod arm, and BaseSchema is .passthrough() with a [key: string]: any index signature, so an undeclared key is KEPT, not refused. Enforce was refused per key: <Chatbot>, the component this registration renders, declares none of the six, so enforcing means growing a component prop or publishing a second spelling of a key that already works. showAvatars is the one key the FENCE turned dark rather than a key nothing ever read: <ChatbotEnhanced> has such a prop and chatbot-floating's raw props spread delivered an authored value to it until #7708 ruled fence (PR #8077). The distinction is recorded in the tombstone comment, the changeset and the pin. processVisibility is NOT folded in — chatbot-enhanced reads it (0/1/0) — and is pinned live as the scope control. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01YBWFb5YgMU5dw8p2VKj16S
github-merge-queue Bot
pushed a commit
that referenced
this pull request
Sep 6, 2026
…es (#8154) loading, showAvatars, userAvatar, assistantAvatar, markdown and height were declared on ChatbotSchema, mirrored on its Zod twin, and read by no plugin-chatbot registration: a schema.KEY census per ComponentRegistry.register body returns 0/0/0 for all six, with placeholder 1/1/1, messages 1/1/1, userAvatarUrl 1/1/1, maxHeight 1/1/0, floatingConfig 0/0/1 and processVisibility 0/1/0 lit on the same instrument. Each becomes `?: never` on complex.ts plus retirementTombstone() on complex.zod.ts — both halves, the convention #6972 / #6355 / #7779 already carry. Deleting them was the wrong route: all six have a Zod arm, and BaseSchema is .passthrough() with a [key: string]: any index signature, so an undeclared key is KEPT, not refused. Enforce was refused per key: <Chatbot>, the component this registration renders, declares none of the six, so enforcing means growing a component prop or publishing a second spelling of a key that already works. showAvatars is the one key the FENCE turned dark rather than a key nothing ever read: <ChatbotEnhanced> has such a prop and chatbot-floating's raw props spread delivered an authored value to it until #7708 ruled fence (PR #8077). The distinction is recorded in the tombstone comment, the changeset and the pin. processVisibility is NOT folded in — chatbot-enhanced reads it (0/1/0) — and is pinned live as the scope control. Claude-Session: https://claude.ai/code/session_01YBWFb5YgMU5dw8p2VKj16S Co-authored-by: Claude <noreply@anthropic.com>
This was referenced Sep 24, 2026
akarma-synetal
pushed a commit
to akarma-synetal/objectui
that referenced
this pull request
Sep 28, 2026
…ts name 41 objectui issues that answer 404 (objectui#10803, batch 5) (objectstack-ai#10892) Part of objectstack-ai#10803 Clause-②: no Dispatched implementation of the `domain:ui` seat 2 claim (comment `5862515185`) on objectui#10803, batch 5, session `https://claude.ai/code/session_014mXUNuFomfj24w7s1pZzhN`. Citations only: no sentence's claim moves, and every edited pending changeset's frontmatter is byte-identical. The runtime text that moves loses its dead pointer and nothing else, all in `@object-ui/types`: the six `ChatbotSchema` tombstone guidance strings open `RETIRED (ADR-0049) —` (rulings `5861129870` and `5861311219`), the `page` node's `actions` refusal, an ADR-0049 tombstone too, cites `(ADR-0049)` in place of its dead card (ruling `5861129870`, as measured in patch round 1), and the `ChartDataSeriesSchema.variant` description drops its pointer (amendment `5860244997`, Q1 = A). Two tests that pin changed text follow it, anchor only (ruling `5861900779`, route a). **Patch round 1** answers the contract review's one FAIL item on `afed8aa6b`: Special cases 7 had read the `actions` refusal as not ADR-0049 from a proxy (the `breadcrumbs` changeset, which only says the 2026-09-09 RULING is not borrowed). Measured from the retirement itself, it is an ADR-0049 tombstone, so under the seat's option (a) its string now reads "never was (ADR-0049): no renderer reads it". The runtime-strings changeset follows, and the sweep changeset notes that `@object-ui/types`' runtime text is declared separately. One text-only commit, `8255f14f4`; no merge, no force-push. ## Why `Part of`, not a closing line The card stays open for the other 23 numbers of the family remainder (seat amendment `5860244997`, Q2 = A), listed in **Acceptance notes** 1, and for the bare sister-repo numbers the C0 census below lists. ## Premise, re-measured - REST `GET /repos/objectstack-ai/objectui/issues/N` for each of the 41: all 41 answer 404, and a second read of each answers 404 again (41 of 41). `GET .../pulls/N` answers 404 for all 41 too. Lit controls: objectui#10533 and objectui#7714 answer 200. No number was dropped from the batch. - The same 41 in objectstack all answer 200. Three of them are the sites' real subject, read not guessed, and are re-qualified (Zone 1 item 5): - objectstack#3720 "列表行级 edit/delete 与批量删除未接入服务端 effective 操作集": the five `[objectstack-ai#3720]` sites in `ObjectGrid.tsx`, `rowCrudAffordances.ts` and `ListView.tsx` describe that gate, and their landing `ba4514511` is titled "gate list row Edit/Delete and bulk delete on the effective operation set (objectstack#3720)". - objectstack#5506, the "five more hardcoded English chrome strings" card: `useDetailTranslation.ts` wrote "the SAME key objectstack-ai#5506 gave NavigationOverlay's identical handle", and nine sibling in-scope lines already cite `objectstack#5506` for those strings. - objectstack#6936, the metadata-admin predicate fail-CLOSED card: `predicate.ts` wrote "objectstack-ai#6936's `true` verdict" one line below "objectstack#6936's unresolved-path ruling". - Every other site names an objectui card or pull request, confirmed by its landing commit below (for example objectstack#7853 is a `driver-memory` finding, while objectui#7853 is the cite-by-content ruling AGENTS.md objectstack-ai#11 names). - Every edited changeset is pending: it is present in `.changeset/` on `main`. - Branch point: `b2683a2c0`. `main` is now `4f8b7f86a`; its census still reads 215, and the merge of this head with it reads 0. Not a shallow checkout. ## C0: the bare-number remainder (measurement only) The instrument of PR objectui#10875, at the branch point: ``` git grep -hoP '(?:^|(?<=[^\w#&/.\-]))#\d+(?![0-9A-Za-z_])' b2683a2 -- '.changeset/*.md' 'packages/*/src/**' ':!**/__tests__/**' ':!**/*.test.*' ':!**/*.spec.*' ':!**/test/**' ':!**/tests/**' | sort -u | wc -l ``` - **998 distinct bare numbers** at `b2683a2c0` (1002 at batch 4's `e32dae160`): the eight batch 4 re-pointed are gone, and four are new since: objectstack-ai#10321, objectstack-ai#8365 and objectstack-ai#9242 (at or below 10900, each read once: 200) and #20051. - **Above 10900: 23**, batch 4's 22 plus #20051. Each was read once with REST `GET .../issues/N`: **23 answer 404**, and each answered 404 again on a second read (23 of 23); `GET .../pulls/N` answers 404 for the 18 that are not colours. - None of them is an objectui citation, and none sits in a sentence or hunk this batch edits, so **nothing is folded here**: - **5 are CSS colours:** #334155, #336699, #475569, #737373, #953800. - **18 are objectstack numbers**, 16 of which answer 200 there with the sites' own subject. #13033 and #13413 answer 404 in objectstack too. Sites (lines / files, at `b2683a2c0`) are listed in **Acceptance notes** 2 for later batches. ## Census (the enumeration pin for this batch) The instrument PR objectui#10854 printed and PRs objectui#10869 / objectstack-ai#10875 reused, with this batch's 41 numbers substituted (REF = a commit or tree): ``` git grep -nE '(objectui#|#|issues/)(7091|7097|7108|7177|7612|7620|7623|7658|7666|7667|7669|7678|7681|7682|7703|7704|7708|7804|7844|7853|7869|7874|7877|7926|7959|7967|7979|7980|8058|8060|3720|5420|5503|5506|5737|6467|6936|6945|7622|7662|7684)([^0-9]|$)' REF -- '.changeset/*.md' 'packages/*/src/**' ':!**/__tests__/**' ':!**/*.test.*' ':!**/*.spec.*' ':!**/test/**' ':!**/tests/**' | grep -v 'objectstack#' | wc -l ``` - REF = `b2683a2c0` (branch point, and `main` now): **215** lines, 74 changeset lines in 55 files and 141 src lines in 47 files. - REF = `8255f14f4` (this head): **0**. Round 0's head `afed8aa6b` also read 0. - This head merged with a fresh `main` (`4f8b7f86a`, `git merge-tree --write-tree`, clean, tree `08f80168a`): **0**. REF = `4f8b7f86a` alone: 215. - **The filter hid nothing this time.** The 18 branch-point lines the `grep -v 'objectstack#'` drops name only objectstack#6936 or objectstack#5506; with those tokens deleted, none of them matches the instrument (0). Unfiltered, this head reads **26**, all live `objectstack#` lines: the 18, the seven re-qualified here, and the same spelling in the new sweep changeset. - Lit control, the same printed instrument over live objectui#7714 at this head: 17 lines. Hex-colour false positives (a number followed by a hex letter) at the branch point: 0. - **Out of scope, as it stands** (the instrument with the 41 numbers, unfiltered, whole tree at this head): 367 test lines in 98 files, 23 scripts lines in 5 files, 0 `.github`, 5 governed lines in 4 files (`AGENTS.md` and three `skills/objectui` files), and 51 other lines in 18 files: published `CHANGELOG.md` history (21 lines in 8 files), `content/docs` (11 in 4), `apps/console` (7 in 3), `eslint-rules` (8 in 2) and `eslint.config.js` (4). See **Acceptance notes** 4. ## Citation form - The 9-character backticked sha of the commit on `main` that landed the change the sentence rests on, as in PRs objectui#10707 / objectstack-ai#10766 / objectstack-ai#10797 / objectstack-ai#10854 / objectstack-ai#10869 / objectstack-ai#10875. All 40 distinct shas in the mapping are ancestors of `main`: `git merge-base --is-ancestor`, exit 0 each. Control leg in the same checkout: the head of PR objectui#10714 (`d2afdb6bf`) answers exit 1, and a known ancestor, `5f789538d`, answers exit 0. `git rev-parse --short=9` returns the same 9 characters for each. - A changeset's pointer to its OWN card is dropped, not replaced: `@changesets/changelog-git` prefixes each released entry with the hash of the commit that added the file, which is that landing. Where dropping it left a dangling word, the clause was minimally repaired (**Special cases** 5). - Where the sentence rests on a RULING, the ruling is cited by its date (**Special cases** 4). - Where a dead card sits beside a live pull request that IS its landing, the dead card is dropped and the live pointer stays (**Special cases** 1). - Where the claim lived only on the dead card (a finding, a ruling, a scope, a comment), the sha only locates it: "the card behind SHA", "the ruling behind SHA" (**Special cases** 3). Review `5861306588`'s lesson: a sha must carry what the sentence cites it for. - Where nothing answers, the pointer is dropped and the sentence names the card by role (objectstack-ai#7612, objectstack-ai#7666, objectstack-ai#7682, objectstack-ai#7844, and the open objectstack-ai#7804 parent). - Runtime text carries no sha: see **Special cases** 7. ## Mapping (number to resolution) Lines / files are the branch-point census for that number. "Method" is how the landing was found. The last column is why that commit carries what the sentence cites it for. | dead number | resolution | method | lines / files | why the commit carries it | |:--|:--|:--|:--|:--| | objectstack-ai#7091 (a pull request) | `969ba84f4` | `git log --grep`: its squash subject ends "(objectstack-ai#7091)" | 1 / 1 | it registers the console `app:launcher` renderer, which is what made the "not a page" clause false | | objectstack-ai#7097 | `3cab5703b`; own-card pointer dropped | changeset's adding commit | 3 / 3 | it adds the `2xl` read arm and the literal `GRID_COLS_2XL` class map, the value-level fix both sentences name | | objectstack-ai#7108 | `1267508ea`; own-card pointer dropped | changeset's adding commit | 3 / 2 | it tombstones a deleted record reference in the approvals inbox, the platform-flagged class the sentences name | | objectstack-ai#7177 | own-card pointer dropped (landing `ebc05b4d6`) | changeset's adding commit | 1 / 1 | the only site is its own changeset's pointer | | objectstack-ai#7612 | "the 2026-09-20 ruling"; the card by role | `git log --grep` finds only `2ca238075`, the objectui#10150 record of the ruling; the switch, PR objectui#10054, closed unmerged | 3 / 2 | nothing landed: the ruling was "no". See **Special cases** 3 and 4 | | objectstack-ai#7620 | `c15d7eca6`; "ruling A behind `c15d7eca6`" where the sentence cites the ruling; own-card pointer dropped | changeset's adding commit, `git log -S` | 13 / 10 | it moves `data-table`'s date-only fallback onto `formatDate`'s default face, and its message records the two-faces fact pattern the sentences cite | | objectstack-ai#7623 | `5d0876c5c`; own-card pointer dropped | changeset's adding commit | 3 / 3 | it retires the inert dashboard-root `title` declaration | | objectstack-ai#7658 | `90c6d090d`; own-card pointer dropped | changeset's adding commit | 2 / 2 | it keeps code-span content out of `extractToc`'s HTML strip | | objectstack-ai#7666 | nothing answers: the card by role ("already-filed defect") | `git log -S` finds only `a472b0716`, which records it as an already-filed defect | 1 / 1 | see **Special cases** 3 | | objectstack-ai#7667 | `a472b0716`; own-card pointer dropped | changeset's adding commit | 2 / 2 | it gives `extractToc`'s emphasis rules the flanking rule that keeps `SCREAMING_SNAKE` underscores | | objectstack-ai#7669 (a pull request) | `a3eb5d07a` | `git log --grep`: its squash subject ends "(objectstack-ai#7669)" | 2 / 2 | it retires `FloatingChatbotConfig.triggerIcon` as the tombstone both sentences name | | objectstack-ai#7678 | `5f8190c8c` ("in its amended form"); own-card pointers dropped | changeset's adding commit; its body opens "Part of objectstack-ai#7678" | 21 / 14 | it brings every statement of the retire-vs-remove discriminator to the amended form, the surviving-carrier precondition included. See **Special cases** 6 | | objectstack-ai#7681 | `01c27c431`; own-card pointer dropped | changeset's adding commit | 6 / 4 | it routes a `dataKey`-shaped series through `normalizeSeries`, removing the fast path, so its `type` override is honoured | | objectstack-ai#7682 | nothing answers: "a separate card"; runtime pointer dropped | `git log -S` finds only `8fe8e5c16`, which defers that decision to it | 3 / 3 | see **Special cases** 3 | | objectstack-ai#7703 | `a4611b3e2`; runtime text `RETIRED (ADR-0049) —`; own-card pointer dropped | changeset's adding commit | 15 / 3 | it retires the six dark `ChatbotSchema` keys as ADR-0049 tombstones | | objectstack-ai#7704 | `20316bac3`; "the card behind `20316bac3`" for the scope note; own-card pointer dropped | changeset's adding commit | 3 / 2 | it adds `chatbot-enhanced` and `chatbot-floating` to `SchemaRegistry` | | objectstack-ai#7708 | `d3499b315`; PR objectstack-ai#8077 kept where it stands beside the dead card; "the card behind `d3499b315`" for the ruling; own-card and runtime pointers dropped | changeset's adding commit, `git log --grep` | 13 / 6 | it fences `chatbot-floating`'s raw props spread through `toDomProps` at the head of the element | | objectstack-ai#7804 (the handler-key parent) | per slice, by `git blame`: `75fca9669` (data-table), `7ca6ddd4b` (detail), `f1cd29032` (list-view), `5a41ce733` (object-kanban), `8d50bc2bf` (objectql.ts), `604476d97` (tree-view); own-card pointers dropped; "the handler-key parent card" where the sentence means the open parent | each slice changeset's adding commit, and `git blame` per line | 49 / 17 | each commit declares exactly the handler keys the site's arm or face names | | objectstack-ai#7844 | nothing answers: "their own card" | `git log -S` finds only `c4a1d39fe`, which leaves both sub-shapes dark | 1 / 1 | see **Special cases** 3 | | objectstack-ai#7853 | "the card behind `fa7d66c45`" | AGENTS.md objectstack-ai#11 names `fa7d66c45` as its landing | 3 / 3 | the ruling lived on the card; `fa7d66c45` is its one-site landing | | objectstack-ai#7869 | "the finding objectui#8344 discharged" / "the finding this card discharges" (landing `841dd2b7e`) | `841dd2b7e`'s message; the live objectui#8344's body, "Discharges the finding objectstack-ai#7869" | 2 / 2 | the measurement lived on the dead finding, and the live card that discharged it restates it | | objectstack-ai#7874 | `d9580f464`; PR objectstack-ai#7887 kept beside the dead card; own-card pointer dropped | changeset's adding commit | 3 / 3 | it retires the five dead `timeline.relative.*` rows | | objectstack-ai#7877 | `7dcda9cd4` | `git log -S` on the heading | 1 / 1 | it moves the hand-rolled-table data into `hand-rolled-tables.json`, the heading's subject | | objectstack-ai#7926 | `12b599219`; "the 2026-09-09 ruling"; "the `actions` refusal" by role beside PR objectstack-ai#8870; runtime text `(ADR-0049)`; own-card pointer dropped | changeset's adding commit | 15 / 6 | it refuses `actions` by name on the `page` node, measures `breadcrumbs` at 1 site and leaves it parsing, and pins that | | objectstack-ai#7959 | `36fc74629`; "the card behind `36fc74629`" for its scope restraint | the tests it added are headed "objectui#7959" | 8 / 6 | it adds `readEnvelopeFailureText`, the userMessage-first rule on `PackagesPage`, and the load-failure testid | | objectstack-ai#7967 | own-card pointer dropped (landing `0544ae958`) | changeset's adding commit | 1 / 1 | the only site is its own changeset's pointer | | objectstack-ai#7979 | own-card pointer dropped (landing `bfa4fe728`) | changeset's adding commit | 1 / 1 | the only site is its own changeset's pointer | | objectstack-ai#7980 | `a810bb2ae`; "the card behind `a810bb2ae`" for its surface and its comment; own-card pointer dropped | changeset's adding commit | 4 / 3 | it publishes `readEnvelopeFailureText` from the package entry | | objectstack-ai#8058 | `3f4b45886` | `git log -S storedRelationshipTarget` | 5 / 2 | it adds `storedRelationshipTarget`, the read door under both spellings, and rewrites the "costs nothing" claim | | objectstack-ai#8060 | `7dc31bb8d`; "step 1 of the card behind `7dc31bb8d`" | `git log --grep`: its subject names "(objectstack-ai#8060)" | 8 / 3 | it carries a stored type the designer cannot author through verbatim (`partitionStoredFields`) instead of flattening it to `text` | | objectstack-ai#3720 (C0) | re-qualified `objectstack#3720` | `ba4514511`'s subject; objectstack#3720's subject | 5 / 3 | see **Premise** | | objectstack-ai#5420 (C0) | own-card pointer dropped (landing `b470e9150`) | the adding commit's subject names "(objectstack-ai#5420)" | 1 / 1 | the only site is its own changeset's heading | | objectstack-ai#5503 (C0, a pull request) | `6f017e99c` | its squash subject ends "(objectstack-ai#5503)" | 1 / 1 | it puts the chart min-height floor on the element Recharts measures | | objectstack-ai#5506 (C0) | re-qualified `objectstack#5506` | nine sibling in-scope lines already cite it for the same strings | 1 / 1 | see **Premise** | | objectstack-ai#5737 (C0, "PR objectstack-ai#5737") | `d1ab06f0f` | its squash subject ends "(objectstack-ai#5737)" | 4 / 4 | it declares `record.*` the row-predicate canon and warns on the two deprecated spellings, which is Phase 1 | | objectstack-ai#6467 (C0) | own-card pointer dropped (landing `5ad0641e0`) | its subject names "(objectstack-ai#6467)" before its own "(objectstack-ai#7332)" | 1 / 1 | the only site is its own changeset's pointer | | objectstack-ai#6936 (C0) | re-qualified `objectstack#6936` | the line above already cites it | 1 / 1 | see **Premise** | | objectstack-ai#6945 (C0, "PR objectstack-ai#6945") | `2c45966ff` | its squash subject ends "(objectstack-ai#6945)" | 1 / 1 | it declares the 13 renderer-read keys the sentence says objectui#6150 declared first | | objectstack-ai#7622 (C0, "PR objectstack-ai#7622") | `1cca678ba` | its squash subject ends "(objectstack-ai#7622)" | 1 / 1 | it retires the dashboard-root `title` read on all five surfaces | | objectstack-ai#7662 (C0, "PR objectstack-ai#7662") | `bc640ec56` | its squash subject ends "(objectstack-ai#7662)" | 1 / 1 | it is the change to `SchemaRegistry`'s kanban entry the sentence ascribes to objectui#7645 | | objectstack-ai#7684 (C0, "PR objectstack-ai#7684") | "the contract review of `8fe8e5c16`'s pull request" | its squash subject ends "(objectstack-ai#7684)" | 2 / 2 | the review lived on the dead pull request; the sha only locates it | ## Special cases (the judgement calls) 1. **A dead card beside its own live landing.** - `defaults-table-scan.ts`: "rows of objectui#7874 at all (retired in objectstack-ai#7887)" becomes "rows at all (retired in objectstack-ai#7887)". PR objectstack-ai#7887 answers 200 and is `d9580f464`'s pull request. - `7703-chatbot-dark-keys-retired.md`: "**after** objectui#7708's fence landed as PR objectstack-ai#8077" becomes "**after** the fence landed as PR objectstack-ai#8077". In the same file "until objectui#7708 the `chatbot-floating` registration ended…" becomes "until PR objectstack-ai#8077 the…", and the next sentence's "That card was ruled **fence**, not declare, and landed as PR objectstack-ai#8077" becomes "Its card was ruled **fence**, not declare, and landed as that PR", so it keeps an antecedent. - `8871-page-node-refuses-breadcrumbs.md` and `layout.zod.ts`: "corrects objectui#7926's "1 site"", beside "(PR objectstack-ai#8870)", becomes "corrects the `actions` refusal's "1 site"". The changeset's "objectui#7926's `patch` does not transfer here" becomes "The `actions` refusal's `patch` does not transfer here". 2. **The sister-repo numbers.** `[objectstack-ai#3720]` becomes `[objectstack#3720]` at five sites; "objectstack-ai#5506" and "objectstack-ai#6936's" become `objectstack#5506` and `objectstack#6936`'s. See **Premise**. 3. **The card named by role.** - **objectstack-ai#7612**, `clientValidation.ts`: "The switch was then put to the maintainer on objectui#7612" becomes "…on its own card". Nothing landed: the ruling said no, and the switch PR closed unmerged. The verbatim ruling quote 「7612 只需要服务端校验」 stays as the maintainer wrote it; it carries no `#`. - **objectstack-ai#7666**, `7667-toc-emphasis-flanking.md`: "(objectui#7666, a heading `extractToc` lists…)" becomes "(a heading `extractToc` lists…)". The sentence already calls it "a different, already-filed defect". - **objectstack-ai#7682**: "objectui#7682 owns that decision" becomes "a separate card owns that decision" (`7546-chart-series-keys-declared.md`), and "objectui#7682's decision" becomes "a separate card's decision" (`data-display.ts`). - **objectstack-ai#7844**, `8754-…` changeset: "the two further sub-shapes objectui#7844 records" becomes "the two further sub-shapes recorded on their own card". - **objectstack-ai#7804 as the open parent**: "whose disposition is still open on objectui#7804" becomes "…on the handler-key parent card" (`9447-…`); "both are ledgered to objectui#7804, which owns it" becomes "…to the handler-key parent card" (`handler-key-reads-follow-cast-receivers.md`); `crud.zod.ts`'s "objectui#7804's rows are the handler keys" becomes "the handler-key parent card's rows…". - **"The card behind SHA"**, where the sentence rests on something that lived only on the card: - objectstack-ai#7620's ruling A: "following the maintainer's ruling A on objectui#7620" becomes "…ruling A behind `c15d7eca6`" in four `@object-ui/fields` widgets, and `GridField`'s "the split objectstack-ai#7620 ruled on" becomes "the split ruling A settled". `RelatedList.tsx`: "objectui#7443 and objectui#7620 spent two cards" becomes "objectui#7443 and the card behind `c15d7eca6` spent two cards". - objectstack-ai#7708's ruling: "objectui#7708 ruled FENCE" becomes "The card behind `d3499b315` ruled FENCE", and "That ruling was FENCE" after it becomes "The ruling behind it was FENCE" (`complex.ts`). - objectstack-ai#7853: "objectui#7853 ruled…" becomes "The card behind `fa7d66c45` ruled…" in both objectstack-ai#8875 changesets and in `defaults-table-scan.ts`'s list. - objectstack-ai#7959's scope restraint: "objectui#7959 kept/left it off the public entry as scope restraint — that card's file surface was…" becomes "The card behind `36fc74629` …" (`7980-…` changeset, `app-shell` `index.ts`). - objectstack-ai#7980: "see objectui#7980 comment 5583988475" becomes "as the card behind `a810bb2ae` records" (a comment on a dead card resolves nowhere); `PackageFormDialog.tsx`'s "is objectui#7980's surface" becomes "is the surface of the card behind `a810bb2ae`". - objectstack-ai#7704's scope note (`registry.ts`) and objectstack-ai#8060's "step 1" (`MetadataFieldsPage.tsx`) take the same form. - objectstack-ai#7869: the finding card is dead, and the live objectui#8344 discharged it and restates the measurement, so `base.zod.ts` reads "That is the finding objectui#8344 discharged: an off-spec `size`…" and objectstack-ai#8344's own changeset reads "the finding this card discharges measured…". 4. **Rulings by date.** - objectstack-ai#7612: the `AUTHOR_SHAPE_ONLY_TYPES` heading ends "(the 2026-09-20 ruling)"; `10150-author-shape-header-ruling.md` reads "the switch was put to the maintainer, and the 2026-09-20 ruling was option A". Its last sentence, "requires the docblock to name the ruling's card", becomes "…to name the ruling", because the pin it describes now anchors on the ruling (**Special cases** 8). - objectstack-ai#7926: "objectui#7926's maintainer ruling covers `actions`" becomes "The 2026-09-09 maintainer ruling covers `actions`", and "objectui#7926 does not rule on it" becomes "the 2026-09-09 ruling does not cover it" (`layout.zod.ts`). 5. **Own-card pointers that needed a word.** - `7097-grid-2xl-breakpoint.md`: the pointer sat on its own line, so the period moved up with it. - `7678-tombstone-discriminator-agreement.md`: "the site objectui#7678 was filed about" becomes "the site this card was filed about", the file's own wording two lines earlier. - `7804-object-kanban-handler-keys-judged.md`: "its `KNOWN_UNDECLARED_READS` row naming objectui#7804, which stays open and stays the parent" becomes "its `KNOWN_UNDECLARED_READS` row, which names the parent card". - The six `7804-*` slice changesets keep "(the … slice)" once the pointer goes, as batch 4 kept "(slice 2)". 6. **The discriminator form (objectstack-ai#7678).** "in the form objectui#7678 amended it to" becomes "in its amended form (`5f8190c8c`)" at every site; "as amended by objectui#7678" and "amended by objectui#7678" take the same wording; a bare "objectstack-ai#7678" in a precedent list becomes `5f8190c8c`. `mobile.ts`'s "without the precondition objectui#7678 added" becomes "without the precondition the amendment added", whose antecedent is the preamble's "amended form" two sentences up. `9684-discriminator-cited-once.md`: "objectui#7678's amendment had to be applied by hand" becomes "the amendment (`5f8190c8c`) had to be applied by hand", which is what that commit did. 7. **The runtime strings.** Only the listed text moves in each. | file | member | before | after | |:--|:--|:--|:--| | `types/src/zod/complex.zod.ts` | `ChatbotSchema.loading`, `.showAvatars`, `.userAvatar`, `.assistantAvatar`, `.markdown`, `.height` (tombstones) | "RETIRED (objectui#7703, ADR-0049) — …" | "RETIRED (ADR-0049) — …" | | `types/src/zod/complex.zod.ts` | `ChatbotSchema.showAvatars` (tombstone) | "… unfiltered props spread — was fenced by objectui#7708. Delete the key: …" | "… unfiltered props spread — was fenced. Delete the key: …" | | `types/src/zod/layout.zod.ts` | `PageNodeSchema.actions` refusal | "`actions` is not a key of the `page` node and never was (objectui#7926): no renderer reads it, …" | "`actions` is not a key of the `page` node and never was (ADR-0049): no renderer reads it, …" | | `types/src/zod/data-display.zod.ts` | `ChartDataSeriesSchema.variant` `.describe()` | "… the renderer-internal `current` spelling is not a member (objectui#7682)" | "… the renderer-internal `current` spelling is not a member" | **The ADR-0049 condition, measured from the retirement itself** (Zone 1 item 4): the six tombstones already cited `ADR-0049` beside the dead card; the zod comment above them opens "ADR-0049 retirement tombstones"; their landing `a4611b3e2` is titled "retire six dark ChatbotSchema keys as ADR-0049 tombstones"; and the pending `7703-chatbot-dark-keys-retired.md` says "ADR-0049 retirement tombstones". The `actions` refusal is an ADR-0049 tombstone too, measured the same way: its arm is `retirementTombstone(PAGE_ACTIONS_REFUSAL)`, its landing `12b599219` calls it "an ADR-0049 `retirementTombstone` arm", and `7926-page-node-refuses-actions.md` calls it "an ADR-0049 refusal arm". Its string has no `RETIRED` opener, so its own parenthetical becomes `(ADR-0049)`, the spelling its `breadcrumbs` sibling already uses. The TypeScript twins in `complex.ts` read "ADR-0049 RETIREMENT TOMBSTONE — `loading` (`a4611b3e2`)", the comment form. 8. **What follows the text, and nothing else in it** (ruling `5861900779`, route a; ruling `5861129870`). - `chatbot-dark-keys-retired-7703.test.ts` pins the six guidance strings as full literals. Their openings follow to `RETIRED (ADR-0049) — `, the `showAvatars` literal follows "was fenced.", its `prescriptive` anchor `'was fenced by objectui#7708'` becomes `'was fenced.'`, and `toContain('RETIRED (objectui#7703')` becomes `toContain('RETIRED (ADR-0049) —')`. Its test names and `@ts-expect-error` comments still read "RETIRED (objectui#7703)"; they describe the `?: never` TypeScript face and are out of class (**Acceptance notes** 6). - `clientValidation.authorShapeRuling-10150.test.ts` pins that the `AUTHOR_SHAPE_ONLY_TYPES` header names the ruling and `extractDraftBody`. Its subject `'objectui#7612'` becomes `'the 2026-09-20 ruling'`; its message and title are unchanged. - No test, doc or changeset quotes the `actions` refusal or the `variant` description verbatim: a whole-tree search for the old strings returns only the source lines. 9. **The anchor sweep (PR objectui#10875 Special cases 10 shape).** - **Instrument:** every string, template, numeric and regex literal in all 3822 test and script files under `packages/` and `scripts/`, enumerated with `git ls-tree` at head and parsed with TypeScript (113091 distinct literals). - **Test applied to each literal:** does its occurrence count DROP between `b2683a2c0` and head in any file this PR changes, in raw text or in a comment-flattened form? - **Run 1, on `8c267051f`** (the sweep commit, before any test moved): 336 distinct literals drop. Read against the files they sit in, the specific anchors are the two pins in item 8 (the six literals, the `prescriptive` anchor and the `toContain` in the first; `'objectui#7612'` in the second). Also listed and read: the `'objectui#7804'` values of `check-handler-key-read-sites`'s `KNOWN_UNDECLARED_READS` ledger and its test, which assert over the script's own data and never read a changed file, and `tombstone-discriminator-agreement-9684`'s `/RETIRED \(objectui#/`, which it applies to `mobile.ts` alone, untouched here. - **Everything else** is a generic token (whitespace, punctuation, single letters and digits, `objectui#`, character-class regexes of whole-tree scanners), none an assertion over text this PR changed. - **Run 2, on round 0's head `afed8aa6b`:** 327 distinct literals; the nine literals the two pins moved are gone from the drop set, nothing new appears, and the only one left naming a batch number is that ledger value. - **Run 3, patch round 1, over the delta `afed8aa6b..8255f14`:** 18 (literal, file) drops, every one a generic token (`' is'`, `'dropped'`, a `/(\w+):/g` scanner); no test pins the changed refusal text. ## Held **By the serial rule: nothing.** Patch round 1 re-read: 8 open besides this PR (objectui#10896, objectstack-ai#10891, objectstack-ai#10890, objectstack-ai#10889, objectstack-ai#10888, objectstack-ai#10777, objectstack-ai#10278 and the release PR objectstack-ai#5400; objectstack-ai#10886 and objectstack-ai#10714 have left the open list). New since round 0, objectui#10896 shares `plugin-grid/src/ObjectGrid.tsx` and `plugin-list/src/ListView.tsx`; trial merges of this head with every open head are clean except objectstack-ai#10278 (conflicts identical against `main` alone) and the release PR objectstack-ai#5400, which regenerates. The round's one source edit, in `layout.zod.ts`, is held by no open PR. Round 0, re-read before its push: 11 open besides this PR. They are objectui#10890, objectstack-ai#10889, objectstack-ai#10888, objectstack-ai#10886, objectstack-ai#10884, objectstack-ai#10852, objectstack-ai#10780, objectstack-ai#10777, objectstack-ai#10714, objectstack-ai#10278 and the release PR objectstack-ai#5400. - Five of them hold files this PR edits: - objectui#10890: `types/src/ai.ts`; - objectui#10888: `.changeset/8344-node-recursion-point-redirect.md`; - objectui#10886: `types/src/objectql.ts` and `types/src/zod/objectql.zod.ts`; - objectui#10714: `types/src/data-display.ts`, `types/src/layout.ts`, `types/src/zod/data-display.zod.ts` and `types/src/zod/layout.zod.ts`; - objectui#10278: `plugin-grid/src/ObjectGrid.tsx`. - Each one's hunks were read against its merge-base, with 3 lines of context. **None of the 47 lines this PR changes in those files appears inside or beside any of their hunks.** - Trial merges of each open PR head with this head (`git merge-tree --write-tree`) are clean for objectstack-ai#10890, objectstack-ai#10889, objectstack-ai#10888, objectstack-ai#10886, objectstack-ai#10884, objectstack-ai#10852, objectstack-ai#10780, objectstack-ai#10777 and objectstack-ai#10714. - objectstack-ai#10278 (`eab4c8e52`) conflicts in `ObjectGrid.tsx`, `plugin-grid/README.md` and `content/docs/plugins/plugin-grid.mdx`, and conflicts identically against `main` alone. - PR objectstack-ai#5400 (Version Packages) regenerates and is not a hold. ## Changesets - `.changeset/10803-dead-citation-sweep-fifth-batch.md`, EMPTY frontmatter: the comment-only edits in 15 released packages and the private `@object-ui/test-support`. No published behaviour changes through them. It marks `@object-ui/types` as the package "whose runtime text is declared separately below" (patch round 1) and points at the second file for it. - `.changeset/10803-fifth-batch-runtime-strings.md`, `'@object-ui/types': patch`: the six tombstones open `RETIRED (ADR-0049) —` with everything after the dash unchanged except the `showAvatars` pointer; the `actions` tombstone cites `ADR-0049` in place of its dead card; and the `variant` description loses its pointer. No key, path, issue code, accept set, refusal or severity moves. ## Proof of prose-only (C4), against `b2683a2c0`, on this head `8255f14f4` - **Source.** Each of the 49 touched `.ts` / `.tsx` files (47 non-test sources and the two pins) was parsed at `b2683a2c0` and at this head with TypeScript's `createSourceFile` and re-printed by `createPrinter({ removeComments: true })`. - 44 of 49 prints are identical. - The other 5 are equal once exactly the 19 listed substitutions are applied to the base print, each matched once: `complex.zod.ts` 7 (six openings and the fence pointer), `layout.zod.ts` 1 (the pointer becomes `(ADR-0049)`), `data-display.zod.ts` 1, `chatbot-dark-keys-retired-7703.test.ts` 9 (the same seven, the `prescriptive` anchor and the `toContain` anchor), `clientValidation.authorShapeRuling-10150.test.ts` 1. - 0 parse diagnostics. - Lit controls on the same instrument: dropping "(ADR-0049)" from the `loading` string moves the print; re-spacing the "ADR-0049 retirement tombstones" comment does not. - **Changesets.** The frontmatter block of every one of the 55 edited changesets is byte-identical at `b2683a2c0` and this head (55 of 55, by md5). The overwrite gate below agrees. - **Scope of the diff:** 106 files: 55 edited and 2 new changesets, 47 non-test source files, and 2 tests. ## Gates, on this head `8255f14f4` Each line is the gate's own verdict and exit code, captured by redirect-then-`$?`. - `node scripts/check-changeset-presence.mjs`, exit 0: "48 source file(s) of 15 released package(s) changed, and this change declares 2 changeset(s): .changeset/10803-dead-citation-sweep-fifth-batch.md, .changeset/10803-fifth-batch-runtime-strings.md." (48 = the 46 released sources and the 2 tests under released packages' `src/`; `test-support` is under a package changesets ignores.) - `pnpm changeset:check`, exit 0: "All workspace packages are in the changeset fixed group." / "No changeset declares a `major` bump." - `node scripts/check-changeset-overwrite.mjs` (report-only), exit 0: "2 changeset(s) added, 55 modified, 0 deleted". `declared at base` equals `declares now` for 55 of 55. - `pnpm check:changeset-claims` (report-only), exit 0: - born-false: "Every one of those 2 address(es) either names the tree it was read from, or points at a line this change does not move"; - self-contradiction: "Every package declared across those 45 body(ies) is either not negated in its own prose, or negated only in an ASPECT of it that may legitimately hold still"; - the standing notice "103 pending changeset(s) describe a file this change touches". - Read against the diff: a pending changeset quoting a replaced pointer would itself carry the dead number and so sit in the census, which reads 0. No pending changeset quotes a changed runtime string. - `pnpm check:control-bytes`, exit 0: "check-control-bytes: OK (scanned 9145 tracked text file(s); skipped 85 binary)." - `pnpm check:new-line-citations`, exit 0: "VERDICT new-cross-file-line-citations: 0 new citation(s), enforcement report-only -> exit 0". - Also run: `pnpm check:pending-changeset-literals`, exit 0: "No test source names a pending changeset."; the governed-surface predicate over the 106 paths, exit 0: "NOT GOVERNED — 106 path(s) checked against 5 governed surface(s); none matched." (lit control `AGENTS.md`: exit 3). **Tests**, through the shared verify lock. - **Red, reproduced first.** The branch-point copies of the two pins were written over the committed sweep tree and run: `Test Files 2 failed (2)`, `Tests 13 failed | 29 passed (42)`, for example "expected 'RETIRED (ADR-0049) — never read: chat…' to contain 'RETIRED (objectui#7703'". Both files were then restored with `git checkout HEAD --`, each checked by blob hash against `HEAD`, and `git diff HEAD` came back empty. - **Patch round 1, on this head `8255f14f4`:** `page-actions-refusal-7926` (it pins the message's remedy words `actions`, `children` and `page:header` and that it is not zod's generic text, never the pointer; verified by reading it), `page-breadcrumbs-refusal-8871`, `guide-layout-page-buttons-7926`, `zod-mirror-parity`, `chatbot-dark-keys-retired-7703`, `clientValidation.authorShapeRuling-10150` and `tombstone-discriminator-agreement-9684` give `Test Files 7 passed (7)`, `Tests 134 passed (134)`, under `VERDICT command-exit 0`. A whole-tree search for "never was: no renderer" finds no test or doc quoting it (only the source line and the runtime-strings changeset, both updated); an anchor sweep over the round's delta finds only generic tokens. - **Green, round 0, on `afed8aa6b`:** `chatbot-dark-keys-retired-7703`, `clientValidation.authorShapeRuling-10150`, `tombstone-discriminator-agreement-9684`, `page-actions-refusal-7926`, `page-breadcrumbs-refusal-8871`, `chart-series-keys-7546`, `zod-mirror-parity`, `chatbot-display-mode-retired` and `schema-registry-chatbot-keys-7704` give `Test Files 9 passed (9)`, `Tests 173 passed (173)`, under `VERDICT command-exit 0`. **Declared narrowing (per the dispatch: the named checks, not wider sweeps).** NOT MEASURED locally: the full suites and type-check of the 15 touched packages, and eslint. Reason: the comment-stripped syntax tree of 44 of the 49 touched files is identical to `main`, and the other 5 differ only by the listed literal edits. Every test that anchors on a changed string or comment was found by the anchor sweep (**Special cases** 9) and run above. CI runs the full farm. ## Acceptance notes 1. **The remainder rides this card, unchanged (seat amendment `5860244997`, Q2 = A).** Of the 53 numbers PR objectui#10875's Acceptance notes 1 lists, this PR carries the first 30 in ascending order, plus the 11 C0 404s of its Acceptance notes 2. The other **23** are: objectstack-ai#8072 objectstack-ai#8127 objectstack-ai#8137 objectstack-ai#8204 objectstack-ai#8229 objectstack-ai#8248 objectstack-ai#8307 objectstack-ai#8408 objectstack-ai#9231 objectstack-ai#9241 objectstack-ai#9244 objectstack-ai#9365 objectstack-ai#9373 objectstack-ai#9375 objectstack-ai#9542 objectstack-ai#9553 objectstack-ai#9585 objectstack-ai#10117 objectstack-ai#10119 objectstack-ai#10120 objectstack-ai#10129 objectstack-ai#10132 #14026. None of them occurs in any hunk of this diff. 2. **The C0 bare numbers above 10900, for later batches** (sites at `b2683a2c0`, lines / files). Each answers 404 as an objectui issue, and none is an objectui citation, so the fix is a re-qualification to objectstack, not a landing sha: - written with a word that names the repository but not in `repo#N` form ("objectstack PR #N", "upstream #N", "framework #N", "spec #N"): objectstack-ai#11289 (2 / 2), objectstack-ai#11662 (1 / 1), #12616 (1 / 1), #12718 (1 / 1), #13033 (1 / 1), #13413 (1 / 1), #13733 (1 / 1), #13906 (6 / 3), #14274 (1 / 1), #15469 (2 / 2), #20160 (2 / 2); - written bare: #13337 (1 / 1), #13632 (3 / 3), #13855 (1 / 1), #14945 (1 / 1), #15948 (1 / 1), #17493 (1 / 1), #20051 (1 / 1); - #13033 and #13413 answer 404 in objectstack as well; - not citations: #334155, #336699, #475569, #737373, #953800 (CSS colours). 3. **A live-but-wrong bare number beside a re-qualified one.** `rowCrudAffordances.ts` cites "(`/me/permissions` `apiOperations`, objectstack-ai#3391)" at the two sites that now read `[objectstack#3720]`. objectui#3391 answers 200 with an unrelated subject (a record-header api action placeholder), while objectstack#3391 is the apiMethods whitelist contract card that objectstack#3720's own title names. It is not a 404, so it is outside the family pin and was not touched. Carrier: none. 4. **Published or shipped text outside the two in-scope classes still names some of these numbers** (counts in **Census**). It is untouched by scope, as another class the triage split did not scope here: published `CHANGELOG.md` history (never re-addressed), `content/docs` (for example the chatbot and layout guides), `apps/console` (the approvals inbox and the agent-connect section), `eslint-rules` and `eslint.config.js`, three `skills/objectui` files and `AGENTS.md` (governed). Carrier: none. 5. **Filenames are not citations.** Pending changeset and test FILENAMES carry several of these numbers. They stay, as in PRs objectui#10707, objectstack-ai#10797, objectstack-ai#10854, objectstack-ai#10869 and objectstack-ai#10875. 6. **Test comments left as they are.** `chatbot-dark-keys-retired-7703.test.ts`'s test names and `@ts-expect-error` comments still read "RETIRED (objectui#7703)", and `clientValidation.authorShapeRuling-10150.test.ts`'s header, title and message still name objectui#7612. Only anchors moved (ruling `5861900779`). --- _Generated by [Claude Code](https://claude.ai/code/session_014mXUNuFomfj24w7s1pZzhN)_ --------- Co-authored-by: Claude <noreply@anthropic.com>
This was referenced Sep 29, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes #7708
What
chatbot-floating's registration (packages/plugin-chatbot/src/renderer.tsx) ended its FloatingChatbot element with a rawpropsspread (object-spread, three dots), LAST — where its two siblings (chatbot,chatbot-enhanced) spreadtoDomProps(props)(object-spread) FIRST. Per the card's ruling (triage comment 5550678895 on #7708, PM dispatch comment 5559736018): fence it, matching the siblings. Option 2 (declare the three keys onChatbotFloatingSchema) was ruled out — it would fossilise an accidental channel as contract (AGENTS.md #0.1) and leaves themessagesoverride untouched.This closes all three consequences at once:
messagesseed used to override the livemessagesprop (bound to the runtime messages) on every render (the raw spread landed after it), so neither the user's own message nor anautoResponsereply ever appeared — the p2 half of the card's grade. Fixed.processVisibility,surfaceandshowAvatarsgo dark onchatbot-floatingnodes. These reached the panel's ChatbotEnhanced component unfiltered even thoughChatbotFloatingSchemanever declared them. The three tripwire pins inrenderer.authoring-faces-7655.test.tsx(section 4, from finding(types):ChatbotSchemapinstypeto'chatbot', sochatbot-enhancedandchatbot-floatingnodes have no authoring-face type #7655/PR feat(types): one named authoring-face type per chatbot registration — ChatbotEnhancedSchema and ChatbotFloatingSchema (objectui#7655) #7705) are flipped from lit to dark, deliberately, as that test's own docblock anticipated.displayMode/systemPrompt/modelstop leaking as DOM attributes on the panel root (objectui#4425's leak class, the oneplugin-chatbotregistration that hadn't closed it).systemPrompt/modelare still read normally by name — only the second, unfiltered forward is gone.Falsifiable premise checked before committing to the fence (per dispatch): every member
ChatbotFloatingSchemadeclares is consumed either byuseObjectChat(...)offschemadirectly, or forwarded by name below the spread — none depend on the raw-spread channel. Verified by readingChatbotFloatingSchema's own doc comment (packages/types/src/complex.ts) against the registration body before editing. No fork: fencing widens nothing and narrows only the accidental channel, matching the dispatch's Clause-② = no.Why
Quoting the card: "⛔ Note the asymmetry: only the first option addresses the p2 half." The
messagesoverride is a user-visible functional break (a floating chat cannot show anything the user types), not a hygiene finding, and only fencing fixes it.Tests
packages/plugin-chatbot/src/__tests__/renderer.floating-spread-fence-7708.test.tsx— the fix's own coverage, previously pinned nowhere:autoResponsereply) renders, through the realSchemaRendererhost.systemPrompt/model/displayModeno longer land as DOM attributes on the panel root.packages/plugin-chatbot/src/__tests__/renderer.authoring-faces-7655.test.tsxsection 4 — the three tripwire pins flip from lit to dark onchatbot-floating;chatbot-enhancedcontrol cases unchanged (still lit — it reads these by name).packages/types/src/complex.ts:ChatbotFloatingSchema's doc comment updated to describe the closed channel — no type-shape change.Run locally, from the repo root:
pnpm exec vitest run packages/plugin-chatbot/— 38 files / 458 tests passed.pnpm exec vitest run packages/types/— 131 files / 2418 tests passed.pnpm --filter @object-ui/plugin-chatbot type-check— pass.pnpm --filter @object-ui/types type-check— pass.node scripts/check-handler-key-read-sites.mjs— pass.node scripts/check-control-bytes.mjs— pass.node scripts/check-published-tsconfig-tooling-exclude.mjs— pass.node scripts/check-changeset-fixed.mjs,check-changeset-no-major.mjs,check-changeset-presence.mjs— pass.eslinton every changed file — 0 errors (pre-existinganywarnings only, none introduced by this diff).Not run locally, both release-time/nightly gates per their own file headers (require a full-repo build the CI-per-PR path deliberately skips):
check:published-distandcheck:sdui-registration-pins(the latter also unrelated in substance — nosideEffectsarray or registration key changed here).Risk / rollback
Behavior change on a shipped registration, shipped
minorper this repo's changeset policy (breaking changes shipminorhere, notmajor— see.changeset/7708-chatbot-floating-spread-fence.md). A document that authoredprocessVisibility/surface/showAvatarson achatbot-floatingnode to rely on this accidental channel loses that effect; author them on achatbot-enhancednode instead. Rollback is reverting this PR (the raw spread returns; no data migration involved).Generated by Claude Code