@azure-id/orc 1.4.0 → 1.4.2

package/CHANGELOG.md

@@ -10,6 +10,235 @@ Format: `### v<version> — <title> _(<date>)_`.
 
 ---
 
+### v1.4.2 - the panel that stops reloading, and the fields you could not read back _(2026-09-05)_
+
+**Still on the unscoped `orc` package?** Do this once first - your `orc upgrade`
+is the pre-v0.56.0 one and cannot install itself. Full detail in the CAUTION at
+the top of this file.
+
+- **Step 1 - release the command from the old package:** `npm uninstall -g orc`
+- **Step 2 - install the current package:** `npm i -g @azure-id/orc`
+- **Step 3 - re-apply it to your project:** `orc update`
+
+**Do not use `npm i -g -f`.** Full detail in v0.56.0 below.
+
+Four more defects in `orc ui` ▸ **CLI Hook Interface**, reported by the person
+who uses it. Three of them turned out to be **the same defect wearing three
+faces**, which is why the fix is one idea rather than four patches.
+
+**STAGING A CHANGE RELOADED THE WHOLE PANEL, AND THAT IS WHY EDITING LOOKED
+BROKEN.** Every staged op re-entered the router: the panel was thrown away, four
+endpoints were refetched, and everything was rebuilt from a skeleton. Move a
+chip, pick a colour, type a name - the page visibly reloaded each time. That is
+bad on its own, and it produced two worse things.
+
+The first is that **a staged change looked like an applied one.** The picture came
+back looking different, so there was nothing left to tell you the write had not
+happened. The second is that **the open editor went stale.** The dialog stayed on
+screen while the panel behind it was replaced, so every control in it was still
+showing the value from BEFORE the change you had just made. You picked a shape,
+the sample did not move; you typed a name, the box did not keep it. The only
+reasonable conclusion was that editing did not work.
+
+Nothing is fetched now until something actually moves the disk - Apply, a
+preset, the reset, the switch, a board change, a recompile. A staged op paints
+from the cache it already has: **zero requests**, the scroll position kept by
+hand. The open editor is registered for that repaint and re-finds its part by
+its stable ref, so what is on screen is always the effective value - **and the
+caret survives**, because an editor that rebuilds on every keystroke and drops
+focus after one letter is not an editor. A name now stages per keystroke rather
+than on blur, which is what made a name typed and then clicked away from race
+the close and sometimes never stage at all.
+
+**AND THE FIELDS THEMSELVES COULD NOT BE READ BACK: `--json is not a summary`,
+found again.** `orc statusline show --json` emitted **twelve** of the
+twenty-four fields a part can carry. So `case`, `prefix`, `suffix`, `glyphs`,
+`format`, `compact`, `min_width`, `precision`, `width`, `min_cols`, `max_cols`
+and `priority` were written to disk correctly by a CLI that has always accepted
+them - and were then **invisible to the panel that wrote them**. Reopen the
+editor and the control was blank. Colour worked; almost nothing else appeared to.
+Every resolved field ships now, plus `authored` - the raw item as the user wrote
+it - so a control can finally say **which values are yours** rather than
+inherited from the catalogue or the colour set. A dot beside a control name is
+that answer.
+
+**SIX PARTS PER LINE, NOT FIVE**, and the cap moved in ONE place
+(`SL_MAX_PER_LINE`), because a slot limit spelled out at each of its five call
+sites is a limit that drifts on the first change. The limit is about reading
+rather than rendering: a line is scanned at a glance, and six is where people
+stopped asking for one more. A spacer, a divider and a fill still do not count -
+they are not things the line SAYS.
+
+**THE DROPDOWN NOBODY COULD READ WAS A PANEL WITH NO COLOURS AT ALL.**
+`hookui.css` had been written against `--bg-sunken`, `--bg-raised`, `--fg` and
+`--fg-dim`, and **none of those tokens exist** - the real names are
+`--surface-2`, `--surface-3`, `--text` and `--text-dim`. Every one of those
+declarations was silently dropped by the browser, which is why the "Between
+parts" select had no background of its own and its open list rendered near-white
+on near-white. All 34 are mapped onto real tokens, and the option list now
+states BOTH halves plus `color-scheme`, because a native popup is drawn by the
+platform and inherits only the half you leave to chance.
+
+**THE PREVIEW IS DRAWN AS A TERMINAL NOW**, because that is the claim it is
+making. A status line in a bare grey box is a string; the same string inside a
+window of the stated width, on a terminal ground, with a **column ruler** under
+it, is a picture of the thing. The bytes did not change - it is still
+`templates/hooks/orc-statusline-render.js`, the hook's own engine - what changed
+is that you can believe it. Each line's cell count sits under the window it
+belongs to instead of in a row of chips a long way from it.
+
+**And you can now see what you are choosing.** `orc statusline preview` takes
+`--theme` and `--glyphs` as a **render-only** override: the same saved layout,
+drawn under a different colour set or symbol set, **writing nothing**. The panel
+uses it for one drawer - *See it in other setups* - holding the three strippings
+(no colour, plain text, no motion), all four colour sets and all seven symbol
+sets. Picking a colour set from its name is a guess; picking it from its picture
+is a decision. The drawer is **closed by default and its rows are fetched only
+when it is opened**: eleven extra renders nobody asked to see is eleven
+subprocesses nobody asked to pay for.
+
+**AND THE SIXTH PART STILL RESET THE BAR TO THE DEFAULT, BECAUSE THE CAP WAS IN
+THREE PLACES.** Raising it in the validator was not enough. The hook has a cheap
+shape guard of its own — it re-checks the compiled file rather than trusting it,
+because a hand-edited compiled file is a file nobody validated — and that guard
+held its own `5`. So a legal six-part line fell straight through to the shipped
+two lines, with the reason recorded in a ledger file nobody was looking at. It
+also counted **every** item, including spacers and fills, which the validator
+never has: a legal five plus a fill fell back too, and always had.
+
+A hook cannot call this CLI — a status line re-renders on every keystroke — so
+the cap now travels ON the lock file (already gated on `orc_version`, so a stale
+copy cannot outlive its build) and the compiler marks each structural item,
+because a compiled op carries an instance id and never a component type. Both
+hooks read both. One cap, one counting rule, three readers.
+
+**AND A NAME YOU TYPED SOMETIMES WENT NOWHERE, WHICH WAS TRUE AND UNSAID.**
+Seven of the thirty-five shapes draw a name: `plain`, `label-value`, `bracket`,
+`angle`, `badge`, `pill`, `stack`. The other twenty-eight — every bar, every
+icon, every bare value — ignore it entirely, and that is the design: a bar with
+a word in front of it is a different shape. But the Name box was offered on all
+thirty-five. It took the text, the CLI wrote it to disk correctly, and nothing
+ever appeared on the bar or in the preview. The same was true of the number and
+text settings on a bar, which lowers a width and a colour ramp and no format at
+all.
+
+Each renderer now publishes the item fields it actually **uses**, and the editor
+greys out the ones this shape ignores — **keeping the slot**, with the reason and
+the list of shapes that do use it, because "this shape cannot use it" and "we
+forgot to offer it" must not look the same, and a value already set has to stay
+readable. The sentence sits under the shape gallery too, since the shape is what
+decides it. A golden test reads the compiler's own switch and fails if the two
+lists disagree.
+
+**AND IT STILL BLINKED, WITH THE NETWORK TAB EMPTY.** Removing the refetch was
+half the fix. `hkPaint` still replaces every child of the panel, and the
+stylesheet fades and slides each one in on a 30ms stagger — so a staged change
+still read as a page reload, on every keystroke, exactly as reported. Entrance
+animations are now removed from the second paint onward, and inside the part
+editor on every rebuild. The transitions are untouched: hover, the drop marker
+and the staged edge are about the pointer rather than about arriving.
+
+One more honesty fix rode along: the sample in the part editor is labelled for
+what it is. `previews` is a per-renderer sample the CLI drew from its own
+fixture, so it shows the **shape** a part will take and never the words or the
+numbers it will carry at run time. It used to say it was drawn with your changes
+in it, and for a label it was not.
+
+---
+
+### v1.4.1 - the board you can actually use _(2026-09-05)_
+
+**Still on the unscoped `orc` package?** Do this once first — your `orc upgrade`
+is the pre-v0.56.0 one and cannot install itself. Full detail in the CAUTION at
+the top of this file.
+
+- **Step 1 — release the command from the old package:** `npm uninstall -g orc`
+- **Step 2 — install the current package:** `npm i -g @azure-id/orc`
+- **Step 3 — re-apply it to your project:** `orc update`
+
+**Do not use `npm i -g -f`.** Full detail in v0.56.0 below.
+
+Four defects in `orc ui` ▸ **CLI Hook Interface**, and the revamp they asked
+for. Reported by the person who designed it, which is the useful kind of report:
+*"this is massive, what the heck is this UI, so plain and not interactive."*
+
+**ADDING THREE PARTS PRODUCED ONE, AND THAT WAS ARITHMETIC, NOT A UI COMPLAINT.**
+Every `+ line N` button in the old palette computed its write's position from the
+**saved** layout rather than from the layout being staged. Three staged adds all
+carried position 1 — and `orc statusline set <line> <pos>` with an item already
+at that position is an **EDIT**, by design. So the second write changed the first
+part's type, the third changed it again, and one part appeared where three were
+asked for. The five-slot cap counted the same stale layout, so it never fired
+either.
+
+The fix is structural rather than careful. A staged change is now a **semantic op
+against a stable ref** — `add` · `remove` · `move` · `set` · `sep` · `doc` — and
+one function, `hkPlan`, replays them in order to produce **both** the board you
+are looking at **and** the writes that produce it, deriving every position at the
+moment that write will run. A second idea of where a part will land is exactly
+the bug this replaces: the board drew one arrangement and the writes made
+another, silently. The cap, the dense-prefix rule and the "line is full" refusal
+all read that same effective board now.
+
+**THE COLOUR SET WROTE ONE LINE AND READ THE WHOLE DOCUMENT.** The picker shelled
+`orc statusline line 1 --theme`, which sets **line 1's** override, while
+`statusline show --json` reports `layout.theme`. So the selected button never
+moved and only a third of the bar changed colour — half working, which is worse
+than not working. New command `orc statusline doc [--theme] [--glyphs] [--ansi]
+[--align-columns]` writes document-level settings, and **clears the per-line
+overrides that would otherwise shadow the choice** — a setting silently shadowed
+is the failure this command exists to fix. One command per scope, and the panel
+calls the one that matches the control.
+
+**THE PREVIEW HAD NEVER HAD COLOUR.** `orc ui` serves under
+`style-src 'self'`, which blocks a parsed `style` **attribute** outright — so
+every SGR colour `ansiBlock` computed was thrown away by the browser, with the
+reason only in a console nobody had open. The properties are assigned through
+CSSOM now, which is not a parse and is not blocked. Loosening the policy to
+`unsafe-inline` for a preview is not a trade worth making.
+
+**AND "WHAT IS THIS THING" NOW HAS AN ANSWER ON THE PAGE.** Every part carries
+its one-line description on the chip, in the picker and in the reference list;
+every control in the editor carries the sentence that says what it does; the
+`80 / 120 / 160` buttons say they are **terminal widths in character cells**;
+and each colour set carries the CLI's own sentence about what it is for.
+
+**The revamp itself, five changes:**
+
+- **The board is the only place anything is applied.** The palette's three
+  `+ line N` buttons are gone. It is a **reference** now — what each part shows
+  and nothing else — because the act of building the bar belongs next to the bar.
+- **Add · Change · Move · Remove are buttons on the chip**, and each opens a
+  **modal that says what it is about to do**. The part picker is one modal, one
+  search box, one click, and it is the only place a part is ever chosen — for a
+  new slot and for swapping an existing one.
+- **Drag and drop**, with the drop edge marked on the chip it will land beside
+  rather than by opening a gap that shoves every other chip sideways while the
+  pointer is still deciding. It is the **shortcut, never the mechanism**: every
+  move has a button and a keyboard equivalent, and drag is switched **off for
+  real** below 600px, where a drop gap stops being a target.
+- **The separator is a dropdown of twelve**, published by the CLI
+  (`separators` in `statusline components --json`) so the panel still names none
+  of them. A value the layout already carries that is not in the set keeps its
+  slot, leads the list and is disabled — the `fixed_executor` rule.
+- **Motion, in `04-motion.css` and nowhere else:** a chip arrives, a chip you
+  just staged pulses three times so you can find it, a picker row slides in as
+  the list narrows under your typing. Every one is FINITE, so the reduced-motion
+  cap leaves each at rest rather than frozen part-way — and the drop marker,
+  the refusal and the staged-edge are borders, never motion.
+
+A refusal is never a shrug: "line 3 is full, it already holds 5 parts" and
+"put something on line 2 first" are two different facts and stay two sentences.
+A line that cannot take a part **says so once** rather than printing the same
+refusal on six identical disabled slots.
+
+`statusline components --json` gains `separators`, `themes_about` and a
+per-component `structural` flag — the panel cannot know which parts do not eat
+one of the five without a second idea of the catalogue, which is the drift this
+panel exists to make impossible.
+
+---
+
 ### v1.4.0 - the agent panel, and the number that was missing _(2026-09-04)_
 
 **Still on the unscoped `orc` package?** Do this once first — your `orc upgrade`

package/README-id.md

@@ -7,13 +7,13 @@
 *Terima permintaan → pahami → rencanakan → beri nilai → kerjakan paralel → periksa → uji → kirim.*
 
 ![npm](https://img.shields.io/npm/v/%40azure-id%2Forc?style=for-the-badge&color=cb3837&logo=npm)
-![Version](https://img.shields.io/badge/version-1.0.0-blue.svg?style=for-the-badge)
+![Version](https://img.shields.io/badge/version-1.4.2-blue.svg?style=for-the-badge)
 ![License](https://img.shields.io/badge/license-MIT-green.svg?style=for-the-badge)
 ![Node](https://img.shields.io/badge/node-%3E%3D18-brightgreen.svg?style=for-the-badge)
 ![Claude Code](https://img.shields.io/badge/Claude_Code-Skills-purple.svg?style=for-the-badge)
 ![Dependencies](https://img.shields.io/badge/dependencies-zero-lightgrey.svg?style=for-the-badge)
 
-**Versi terbaru: v1.0.0** · diperbarui 30-08-2026 · [daftar perubahan lengkap](CHANGELOG.md)
+**Versi terbaru: v1.4.2** · diperbarui 05-09-2026 · [daftar perubahan lengkap](CHANGELOG.md)
 
 **Ada di npm: [`@azure-id/orc`](https://www.npmjs.com/package/@azure-id/orc)** — `npm i -g @azure-id/orc`
 
@@ -619,6 +619,43 @@ Bacalah sebagai catatan putaran itu, bukan sebagai audit terkini:
 **Riwayat lengkap: [CHANGELOG.md](CHANGELOG.md)** — atau `orc changelog`, yang
 hanya mencetak yang lebih baru dari versi yang Anda punya.
 
+### v1.4.2 - panel yang berhenti memuat ulang, dan bidang yang tidak bisa dibaca kembali _(05-09-2026)_
+
+Empat cacat lagi di `orc ui` ▸ **Antarmuka CLI Hook**. Tiga di antaranya adalah
+cacat yang sama dengan tiga wajah berbeda.
+
+**Menyiapkan satu perubahan memuat ulang seluruh panel** - tiap operasi yang
+disiapkan masuk kembali ke router, jadi empat endpoint diambil ulang dan semua
+dibangun lagi dari kerangka. Itu membuat perubahan yang baru disiapkan tampak
+seperti sudah diterapkan, dan meninggalkan editor yang terbuka menampilkan nilai
+sebelum perubahan yang baru saja Anda buat: Anda memilih bentuk, tidak ada yang
+bergerak. Sekarang tidak ada yang diambil sampai sesuatu benar-benar mengubah
+disk; satu operasi yang disiapkan menggambar ulang dari cache dengan **nol
+permintaan**, menjaga posisi gulir Anda, dan editor yang terbuka ikut digambar
+ulang - beserta kursornya.
+
+**Dan bidangnya tidak bisa dibaca kembali.** `orc statusline show --json`
+mengirim dua belas dari dua puluh empat bidang yang bisa dibawa sebuah bagian,
+jadi `case`, `prefix`, `min_cols`, `precision` dan tujuh lainnya ditulis ke disk
+dengan benar lalu tidak terlihat oleh panel yang menulisnya. Sekarang semua
+bidang dikirim, ditambah `authored`, sehingga sebuah kontrol bisa menyebut
+**nilai mana yang milik Anda** dan mana yang warisan.
+
+**Enam bagian per baris, bukan lima** - dipindahkan di satu tempat, dan pemisah
+tetap tidak dihitung. **Menu tarik-turun yang tidak terbaca** ternyata adalah
+panel tanpa warna sama sekali: `hookui.css` ditulis memakai empat token CSS yang
+tidak ada, jadi setiap deklarasi itu dibuang diam-diam oleh peramban.
+
+**Pratinjau kini digambar sebagai terminal** - jendela selebar yang disebutkan
+dengan penggaris kolom - dan `orc statusline preview` menerima `--theme` dan
+`--glyphs` sebagai penimpaan **khusus gambar**, jadi laci baru menampilkan bar
+yang sama di bawah keempat set warna dan ketujuh set simbol. Memilih set warna
+dari namanya adalah tebakan; memilihnya dari gambarnya adalah keputusan.
+
+Detail lengkap di [CHANGELOG.md](CHANGELOG.md).
+
+---
+
 ### v1.3.0 - bangun sendiri baris status Anda _(04-09-2026)_
 
 **Masih memakai paket `orc` yang tanpa awalan?** Lakukan ini sekali dulu —

package/README.md

@@ -7,14 +7,14 @@
 *Intake → analyze → plan → score → parallel subagents → review → verify → ship.*
 
 ![npm](https://img.shields.io/npm/v/%40azure-id%2Forc?style=for-the-badge&color=cb3837&logo=npm)
-![Version](https://img.shields.io/badge/version-1.2.0-blue.svg?style=for-the-badge)
+![Version](https://img.shields.io/badge/version-1.4.2-blue.svg?style=for-the-badge)
 ![License](https://img.shields.io/badge/license-MIT-green.svg?style=for-the-badge)
 ![Node](https://img.shields.io/badge/node-%3E%3D18-brightgreen.svg?style=for-the-badge)
 ![Claude Code](https://img.shields.io/badge/Claude_Code-Skills-purple.svg?style=for-the-badge)
 ![Dependencies](https://img.shields.io/badge/dependencies-zero-lightgrey.svg?style=for-the-badge)
 ![GitHub stars](https://img.shields.io/github/stars/azure-id/orc?style=for-the-badge&color=yellow)
 
-**Latest: v1.4.0** · updated 2026-09-04 · [full changelog](CHANGELOG.md)
+**Latest: v1.4.2** · updated 2026-09-05 · [full changelog](CHANGELOG.md)
 
 **On npm: [`@azure-id/orc`](https://www.npmjs.com/package/@azure-id/orc)** — `npm i -g @azure-id/orc`
 
@@ -575,74 +575,64 @@ a current audit: [EVAL-REPORT.md](EVAL-REPORT.md).
 **Full history: [CHANGELOG.md](CHANGELOG.md)** — or `orc changelog`, which prints
 only what is newer than the version you have.
 
-### v1.4.0 - the agent panel, and the number that was missing _(2026-09-04)_
+### v1.4.2 - the panel that stops reloading, and the fields you could not read back _(2026-09-05)_
 
-**Still on the unscoped `orc` package?** Do this once first — your `orc upgrade`
+**Still on the unscoped `orc` package?** Do this once first - your `orc upgrade`
 is the pre-v0.56.0 one and cannot install itself. Full detail in the CAUTION at
 the top of this file.
 
-- **Step 1 — release the command from the old package:** `npm uninstall -g orc`
-- **Step 2 — install the current package:** `npm i -g @azure-id/orc`
-- **Step 3 — re-apply it to your project:** `orc update`
+- **Step 1 - release the command from the old package:** `npm uninstall -g orc`
+- **Step 2 - install the current package:** `npm i -g @azure-id/orc`
+- **Step 3 - re-apply it to your project:** `orc update`
 
 **Do not use `npm i -g -f`.** Full detail in v0.56.0 below.
 
-The second board. Claude Code renders a custom row for every subagent in the
-agent panel, and that surface is ORC's exact domain: one row per dispatched
-agent, live, while it runs.
-
-```
-● orc-executor-opus-5-low   O5/low   84K   ███▎░░░░  42%   for 17m
-✓ orc-reviewer-opus-5-med   O5/med   31K   █▌░░░░░░  16%   for  4m
-```
-
-**AND IT ANSWERS A QUESTION v1.2.0 SAID COULD NOT BE ANSWERED.** That release
-established that Claude Code records no token usage for a dispatched subagent —
-`isSidechain` is never set, no sidechain message carries a usage block, verified
-across every transcript on two machines — so `orc usage report` has reported
-`tokens: null` for every Claude row ever since, and said why.
-
-That is still true **of the transcript**. It is not true of the agent panel,
-which carries `tokenCount` per task along with the resolved `model` and
-`effort`. So the hook writes down what it is handed, and `orc usage report`
-reads it.
-
-- **It is a FLOOR, and it is labelled one everywhere it appears.** The hook sees
-  a task only while it is in the panel: an agent that started and finished
-  between two renders is never seen, and a count read just before an agent
-  finished is short by whatever came after. `not-seen` means exactly that and
-  is never `0`. A floor reported as a total would be the same class of lie as a
-  zero reported for an unknown.
-- **The record is written even with the board OFF.** It is not part of the
-  display feature — it is a measurement Claude Code hands over either way, and
-  throwing it out because a display setting is off would be the wrong trade by a
-  wide margin. `orc init` and `orc update` wire `subagentStatusLine` for that
-  reason alone, and never clobber one you already have.
-- **A count can only go up**, so a lower reading is a stale one and never
-  overwrites a higher one.
-- **The model and effort are OBSERVED.** ORC's downgrade check has two readings
-  — one derived from the agent's name, one the agent reports about itself. This
-  is the third, and the only one nobody had to be trusted for.
-
-**ONE COMPILER, TWO BOARDS.** The second board reuses the compiler, the render
-program, every shape, every glyph set, the colour model, the validator and the
-gate ladder. What differs is a component set, three filenames and a config key —
-a table, not a fork. A test asserts there is exactly one compiler and that the
-new hook grew no renderer of its own, because that is the cheap mistake this
-whole design exists to avoid.
-
-- **A component belongs to one board**, and the other refuses it by name with
-  the board it belongs to. Two catalogues would be two lists somebody has to
-  keep in step; this is one catalogue with a column.
-- **A subagent row is one line by construction** — Claude Code renders one per
-  task — so the three-line board and its dense-prefix rule simply do not apply.
-- **Three presets**: what the agent is and what it has cost, a watch view with
-  its own context window, and the downgrade check made visible per agent.
-- Every gate rung falls back to **Claude Code's own row**, which is a real
-  answer and a better one than a blank. An empty render hides a task entirely,
-  which is almost never what anybody meant — so it is never emitted.
-
-`subagent_line_custom`, default `off`. Off is Claude Code's row, unchanged.
+Four more defects in `orc ui` ▸ **CLI Hook Interface**. Three were the same
+defect wearing three faces.
+
+**Staging a change reloaded the whole panel** - every staged op re-entered the
+router, so four endpoints were refetched and everything was rebuilt from a
+skeleton. That made a staged change look like an applied one, and it left the
+open editor showing values from before the change you had just made: you picked
+a shape and nothing moved. Nothing is fetched now until something actually moves
+the disk; a staged op repaints from cache with **zero requests**, keeps your
+scroll position, and the open editor repaints with it - caret and all.
+
+**And the fields could not be read back.** `orc statusline show --json` emitted
+twelve of the twenty-four fields a part can carry, so `case`, `prefix`,
+`min_cols`, `precision` and seven more were written to disk correctly and then
+were invisible to the panel that wrote them. Every field ships now, plus
+`authored`, so a control can say **which values are yours** rather than
+inherited. `--json is not a summary`, found again.
+
+**Six parts per line, not five.** **The dropdown nobody could read** turned out
+to be a panel with no colours at all: `hookui.css` was written against four CSS tokens that do not
+exist, so every one of those declarations was silently dropped.
+
+**The preview is drawn as a terminal now** - a window of the stated width with a
+column ruler - and `orc statusline preview` takes `--theme` and `--glyphs` as a
+**render-only** override, so a new drawer shows the same bar under all four
+colour sets and all seven symbol sets. Picking a colour set from its name is a
+guess; picking it from its picture is a decision.
+
+**And the sixth part still reset the bar to the default**, because the cap was in
+three places: the hook re-checks the compiled file with a cheap shape guard of
+its own, and that guard held its own `5` — and counted spacers, which the
+validator never has. The cap now travels on the lock file and the compiler marks
+each structural item, so one cap and one counting rule serve all three readers.
+
+**And a name you typed sometimes went nowhere.** Seven of the thirty-five shapes
+draw a name; the other twenty-eight ignore it, which is the design — a bar with a
+word in front of it is a different shape. The Name box was offered on all
+thirty-five anyway. Each renderer now publishes the fields it **uses**, and the
+editor greys out what this shape ignores, keeping the slot and naming the shapes
+that do.
+
+**And it still blinked with the network tab empty:** `hkPaint` replaces every
+child, and the stylesheet faded each one in on a stagger. Entrance animations are
+off from the second paint onward.
+
+Full detail in [CHANGELOG.md](CHANGELOG.md).
 
 ---