Record that the first step checks every schema rule, and plan moving the writers' schema refusals there

AGENTS.md

@@ -59,7 +59,7 @@ In `docs/decisions.md`:
 - Posts announcing a release wait for its publish, and their drafts wait for the finished release
 - A chunk ruled worse merges where a plan item of its release already restores that code's reading, and that item takes the ruling's places
 - A chunk ruled worse that restructures the places takes a paired ruling after each restructure, against the same base, and merges once one is not worse
-- Writers and carry readers refuse an inline card a link marks
+- The first step checks the schema, and no converter does
 - The gate runs on Deno and Bun
 - The gate installs the tarball
 - Firefox reads the build

docs/decisions.md

@@ -297,7 +297,7 @@ besides hard breaks but no node that takes a link: the panel refused to have
 breaks alone still reads as its destination, then the breaks (§An image reads as an image only alone
 in its paragraph). `adfToLosslessMarkdown` writes a node holding a mark its type does not take as
 `!adf:carry`, so the round-trip holds, except an inline card a link marks, which is refused
-(§Writers and carry readers refuse an inline card a link marks).
+(§The first step checks the schema, and no converter does).
 
 ## A linked image is written `[![alt](url)](href)`, a captioned image as its directive
 
@@ -442,7 +442,7 @@ it joins the second panel's `<a>` to that portable spelling.
 
 HTML nests no `<a>` in an `<a>`, so no element can hold both a card's URL and the target of a link
 marking it. Every writer refuses such a card, and a carry element holding one, as the maintainer
-chose on 2026-11-04 over a new spelling (§Writers and carry readers refuse an inline card a link marks).
+chose on 2026-11-04 over a new spelling (§The first step checks the schema, and no converter does).
 
 A fifth panel (2026-11-04, three readers in the same roles, then four more for what the first three
 split on) was asked how to write a mention, emoji, status, date and inline card, which until then
@@ -749,27 +749,31 @@ stays on `insideRun` alone, since it answers the parted line. A refactor that me
 so the chunk's later rulings read against that merge, the commit `larv-review ruling-base` prints. A restructure past a
 pass's round cap still goes unreviewed, so the PR summary names it.
 
-## Writers and carry readers refuse an inline card a link marks
+## The first step checks the schema, and no converter does
 
-2026-11-04, replacing §No schema validation of 2026-08-23 (its mark refusal 2026-08-25), the
-maintainer. Goal 1. Valid while Atlassian's schema gives a card no link mark.
+2026-11-04, the maintainer. Goal 1 and code health. Valid while Atlassian's schema forbids shapes
+a converter would otherwise meet.
 
-Every writer refuses a document holding an `inlineCard` with a `link` mark, as
-`unsupported-node-shape` at the card's path, wherever the card stands: in a node a writer carries,
-or under a type the library does not know. Every carry reader, `losslessMarkdownToAdf` and
-`htmlToAdf`, refuses a carried node holding one, so no reader builds a document a writer refuses.
-The schema gives a card the `annotation` mark alone (§The node tables answer to Atlassian's
-schema), and spelling the pair would take a new spelling in each format; HTML nests no `<a>` in an
-`<a>`. Another shape joins this entry only at the maintainer's word, once the schema forbids it and every format would
-need a spelling for it. A node type, mark or attribute the schema does not name still rides the
-carry. There is no full schema validation and no exported validator: `isAdfDocument` checks the
-document's shape alone. A refusal that keeps the round-trip is no schema check and stays: a spelled
-node carrying the same mark type twice is refused on its own terms, and input nesting a spelling
-inside its own kind (`*(*a*)*`) names that mark once.
+An Atlassian-schema rule the library applies is checked once, in the first step every call opens
+with: `writableContent` for the writers, `carriedNode` for the carry readers `losslessMarkdownToAdf`
+and `htmlToAdf`. A converter checks none, so a document one call refuses every call refuses, at
+the same path, and what a converter holds is valid ADF. An `inlineCard` with a `link` mark is
+refused there today, as `unsupported-node-shape` at the card's path, wherever it stands, since the
+schema gives a card the `annotation` mark alone (§The node tables answer to Atlassian's schema)
+and HTML nests no `<a>` in an `<a>`. `todo.md` item 167 moves the other refusals a writer makes on a schema
+rule: a node holding a mark type twice, a node holding `text`, or `content` where it takes none,
+and an inline node holding neither.
 
-The earlier entry premised that the site a document is saved to validates it. A hand-built
-document has not reached the site yet, and the viewer/editor persona relies on a refusal arriving
-before the save.
+A rule joins the first step as a refusal only where no format can spell the node whole. A node type,
+mark or attribute the schema does not name, a known node where no section spells it
+(§Unknown nodes ride the carry), and a mark a node cannot hold (§A mark skips a node that cannot
+hold it) ride the carry, which keeps the round-trip. A limit of the engine or the flavour is no
+schema rule and stays where it is met: nesting depth, output length and what a portable line can
+spell. A reader refusing markdown or HTML it cannot read is no check of ADF and stays in the reader.
+
+The earlier entry kept the card refusal alone, premising that the site a document is saved to
+validates the rest. A hand-built document has not reached the site yet, and the viewer/editor
+persona relies on a refusal arriving before the save.
 
 ## The gate runs on Deno and Bun
 

src/adf/document.ts

@@ -147,7 +147,7 @@ export function nodeMarks(node: { marks?: AdfMark[] }): readonly AdfMark[] {
   return node.marks ?? []
 }
 
-// Walks every node, carried or of an unknown type, so a link-marked card is refused wherever it stands (docs/decisions.md §Writers and carry readers refuse an inline card a link marks).
+// Walks every node, carried or of an unknown type, so a link-marked card is refused wherever it stands (docs/decisions.md §The first step checks the schema, and no converter does).
 export function refusedShape(content: readonly AdfNode[]): ConvertError | undefined {
   type Step = { index: number; node: AdfNode; parent: Step | undefined }
   const pending: Step[] = content.map((node, index) => ({ index, node, parent: undefined })).reverse()

todo.md

@@ -6,7 +6,7 @@
 
 `Bar = 0.1`
 
-`Next ID = 167`
+`Next ID = 168`
 
 ## Items
 
@@ -17,6 +17,7 @@
 | 165 | 0.3.0 | 164 |  | **Have a reader panel seated from `## Audience` judge each feature-table row reasonable or not, and file an item for each row it rejects.** | 2 | 2 | 1 | 3 | 3 | 6 by 5 | 7.5 |
 | 132 | 0.3.0 |  |  | **Write tables as `<table>` in `adfToHtml` and read them back in `htmlToAdf`, in place of the carry element.** | 2 | 2 | 2 | 4 | 5 | 6 by 4 | 2.4 |
 | 73 | 0.3.0 | 132, 163, 165 |  | **Announce the package where someone needing an ADF converter already reads: JRACLOUD-77436, the Atlassian developer community and Stack Overflow's ADF-to-markdown questions.** | 1 | 2 | 2 | 3 | 2 | 3 by 8 | 16.9 |
+| 167 | 0.3.0 |  | decision | **Check every Atlassian-schema rule a writer refuses today in `writableContent`, and drop it from the writers.** | 3 | 1 | 2 | 4 | 5 | code health by 5 | 13.1 |
 | 72 | 0.3.0 | 132 | decision | **Re-create `gitea.larvit.se/larvit/adf-codec` as a read-only pull mirror of GitHub, its issues and Actions off and its description pointing at GitHub.** | 1 | 2 | 1 | 2 | 1 | 3 by 1 | 2.5 |
 | 153 | 0.4.0 |  | question | **Decide whether `src/markdown/` root may import `emit/`, `parse/` or `portable/`, and hold the answer in `.oxlintrc.json` and §The source parts by ADF and format.** | 1 | 2 | 2 | 3 | 2 | code health by 3 | 26.2 |
 | 156 | 0.4.0 |  |  | **Stop a leg's container when its script is interrupted, so Ctrl-C ends `ci.sh` or `mutation-tests/run.sh` at once.** | 1 | 2 | 1 | 2 | 1 | code health by 2 | 20.9 |
@@ -26,7 +27,6 @@
 | 155 | 0.4.0 |  |  | **Have `closeBracket` return the link, image or literal bracket it builds, and apply it to the scan in one caller.** | 2 | 1 | 2 | 3 | 4 | code health by 4 | 17.4 |
 | 162 | 0.4.0 |  |  | **Pin each rule of the emphasis sweep with a hand-worked line, so every mutant that lets a pair fall where the one-run-per-pass reference keeps it fails a test.** | 2 | 1 | 1 | 2 | 4 | code health by 4 | 17.4 |
 | 45 | 0.4.0 |  | defect | **Replace `isAdfDocument` with a reader returning `Result<AdfDocument>`.** | 2 | 1 | 3 | 4 | 3 | 1 by 2 | 13.1 |
-| 68 | 0.4.0 |  | defect | **Carry a non-text inline node holding `content` or `text`, a block holding `text`, and a leaf block holding `content`.** | 2 | 1 | 2 | 3 | 3 | 1 by 2 | 13.1 |
 | 127 | 0.4.0 |  | principle | **Hold what `entity-references.test.ts`, `json-text.test.ts` and `result.test.ts` check by asserting outcomes, so none reads source text.** | 2 | 2 | 1 | 3 | 3 | code health by 2 | 13.1 |
 | 159 | 0.4.0 |  |  | **Delete `taskBlocks`'s branch that writes the task marker as its own paragraph, once a test shows no block task item reaches it.** | 1 | 1 | 1 | 1 | 1 | code health by 1 | 10.5 |
 | 144 | 0.4.0 |  |  | **Give every container `docker-runner.sh` and `publish.sh` start a memory limit and an equal swap limit, sized from the largest a gate leg measurably needs.** | 1 | 1 | 1 | 1 | 2 | code health by 1 | 8.7 |
@@ -106,6 +106,30 @@ solves it, then links the GitHub repository, the README's comparison table and t
 opening. The maintainer posts them once `0.3.0` is published (`docs/decisions.md` §Posts
 announcing a release wait for its publish). Found while planning discoverability, 2026-10-04.
 
+### 167. Check every Atlassian-schema rule a writer refuses today in `writableContent`, and drop it from the writers.
+
+The maintainer's call, 2026-11-04 (`docs/decisions.md` §The first step checks the schema, and no
+converter does): the first step holds every schema check, "moved there from wherever else in the
+code they are". It replaces item 68, which would have carried these shapes: a schema-forbidden
+shape is refused once, with its path, in every writer. Move into `refusedShape`:
+
+- `emitLeaf` in `emit/inline-line.ts`: a node holding one mark type twice, and an inline node
+  holding neither `content` nor `text`.
+- `emitDirectiveBlock` in `emit/adf-to-markdown.ts`: a node holding `text`, and a leaf block holding
+  `content`.
+- The HTML and portable writers, which today write the first two without refusing: they refuse them
+  too, which is breaking for a document holding either (`CHANGELOG.md`, `MIGRATION.md`).
+- Check `src/html/` and `src/markdown/portable/` for a refusal of this kind the grep of
+  `unsupported-node-shape` missed, and `carriedNode` for a rule the readers still build.
+
+Stays where it is: a limit of the engine or the flavour (nesting depth, output length, a portable
+task list holding what no list item spells) and a reader's refusal of markdown or HTML it cannot
+read. A node type, mark or attribute the schema does not name, a known node out of place and a mark
+a node cannot hold keep riding the carry. Drop the refused shapes from the README's
+`unsupported-node-shape` causes under writing and from `spec/flavour.md`; the property generators
+build none of them. Lands with items 45, 77 and 80 where it can, since their reader walks every
+node too, but needs none.
+
 ### 72. Re-create `gitea.larvit.se/larvit/adf-codec` as a read-only pull mirror of GitHub, its issues and Actions off and its description pointing at GitHub.
 
 Blocked by every other `0.3.0` item but 73 (the maintainer, 2026-10-05; moved to `0.3.0` 2026-10-28,
@@ -183,14 +207,6 @@ boolean guard is the one export that does not, and it cannot say which branch re
 `not-an-adf-document`'s message already does. Breaking: `MIGRATION.md` shows the guard's
 replacement.
 
-### 68. Carry a non-text inline node holding `content` or `text`, a block holding `text`, and a leaf block holding `content`.
-
-`emitLeaf` (`emit/inline-line.ts`) and `emitDirectiveBlock` (`emit/adf-to-markdown.ts`)
-refuse these with `unsupported-node-shape`, though `isAdfDocument` accepts them and the carry
-round-trips them; `docs/decisions.md` §The code list gives a cause the carry answers no code. Drop
-them from the README's `unsupported-node-shape` cause "Writing, in `adfToLosslessMarkdown`". Found by the README-goals
-audit, 2026-10-04.
-
 ### 127. Hold what `entity-references.test.ts`, `json-text.test.ts` and `result.test.ts` check by asserting outcomes, so none reads source text.
 
 Technical principle "Assert the outcome", which bans scanning the source with `ast` or a regex.