Answer the stability review: guard the argument lookup, and correct the claims the split made false

AGENTS.md

@@ -138,8 +138,8 @@ live Atlassian APIs; property-generated ADF trees; the CommonMark spec suite aga
   fields it claims (`isAdfDocument`); past it everything is typed. Make invalid states
   unrepresentable.
 - `src/adf/` holds ADF's own knowledge and imports no format. Each format directory (`markdown/`,
-  `html/`) parts into `emit/` (ADF→format) and `parse/` (format→ADF), its root holding only what
+  `html/`) parts into `emit/` (ADF→format) and `parse/` (format→ADF), its root holding what both
-  both directions read.
+  directions read.
 - The attribute vocabulary is ADF's: `adf/` walks it and narrows each value to its kind, and a
   format spells the narrowed value. A spelling that re-checks the type is the check's second copy.
 - Explicit over implicit; descriptive names; no catch-all files (`utils`, `helpers`, `misc`); a

src/adf/attribute-vocabulary.ts

@@ -1,7 +1,7 @@
 import type { AdfAttributes } from './document.ts'
 import type { JsonValue } from '../json-value.ts'
 
-export type AttributeKind = 'boolean' | 'json' | 'number' | 'string'
+type AttributeKind = 'boolean' | 'json' | 'number' | 'string'
 
 export type AttributeVocabulary = Readonly<Record<string, AttributeKind>>
 

src/markdown/block-directive-arguments.ts

@@ -1,11 +1,13 @@
 import type { BlockType } from '../adf/block-directives.ts'
 
-const blockArguments: Readonly<Record<string, string>> = {
+const argumentByType = new Map(
+  Object.entries({
     blockTaskItem: 'state',
     panel: 'panelType',
     taskItem: 'state',
-} satisfies Partial<Record<BlockType, string>>
+  } satisfies Partial<Record<BlockType, string>>),
+)
 
 export function blockArgument(type: string): string | undefined {
-  return blockArguments[type]
+  return argumentByType.get(type)
 }

src/markdown/commonmark-grammar.ts

@@ -1,8 +1,5 @@
 export type LinePosition = 'first' | 'later'
 
-const asciiPunctuation = /[!"#$%&'()*+,\-./:;<=>?@[\\\]^_`{|}~]/
-const htmlConstructs = [/^<[!?]/, /^<\/?[A-Za-z][A-Za-z0-9-]*(?:[\s/>]|$)/, /^<[^\s<>@]+@[^\s<>@]+>/]
-
 const controlCharacterRange = '\\u0000-\\u001f\\u007f'
 const autolinkSource = `[A-Za-z][A-Za-z0-9+.-]{1,31}:[^\\s<>${controlCharacterRange}]*`
 const nullCharacterSource = '\\u0000'
@@ -14,7 +11,9 @@ const bracketedAutolink = new RegExp(`^<(?:${autolinkSource})>`)
 const controlCharacter = new RegExp(`[${controlCharacterRange}]`)
 const entityReference = new RegExp(entityReferenceSource)
 const nullCharacter = new RegExp(nullCharacterSource)
+const asciiPunctuation = /[!"#$%&'()*+,\-./:;<=>?@[\\\]^_`{|}~]/
 const firstCharacterOpeners = [/^#{1,6}(?:[ \t]|$)/, /^>/, /^[*+-](?:[ \t]|$)/, /^`{3,}/, /^~{3,}/, /^:{2,}/, /^\|/]
+const htmlConstructs = [/^<[!?]/, /^<\/?[A-Za-z][A-Za-z0-9-]*(?:[\s/>]|$)/, /^<[^\s<>@]+@[^\s<>@]+>/]
 const orderedListOpener = /^(\d{1,9})[.)](?:[ \t]|$)/
 const setextUnderline = /^(?:=+|-+)$/
 const thematicBreak = /^(?:(?:\*[ \t]*){3,}|(?:-[ \t]*){3,}|(?:_[ \t]*){3,})$/

todo.md

@@ -189,7 +189,7 @@ detail is settled at its own milestone.
         HTML will want too, not a markdown spelling.
         **Settled** (the maintainer, 2026-08-27): `markdown/` parts here as well, into `emit/` and
         `parse/` with the shared set at the root — the grammar module, emphasis matching,
-        destination spelling, the tables' markdown halves — and `parse/` arriving with 3b's first
+        the tables' markdown halves — and `parse/` arriving with 3b's first
         file, the rule `html/` already follows. And the node tables, a second copy of
         `spec/flavour.md`'s prose whose mistyped attribute name degrades into a false refusal no
         test catches, get their guard: a test reads the spec's node sections, takes each