Slate v2 TipTap Slate-Candidate Tests Ralplan

status: done score: 0.95 created: 2026-05-11 skill: .agents/skills/slate-ralplan/SKILL.md source harvest: docs/editor-test-harvester/tiptap/report.md

Current Verdict

TipTap does not produce a broad raw Slate test backlog. After stripping Plate-owned/plugin/product rows, the raw Slate result is:

new Slate tests worth considering: 2
already covered raw Slate rows: 5
deferred raw performance row: 1
hard rejects from raw Slate: all TipTap command-chain, extension option, NodeView, mark-range UI, serializer, markdown/input-rule, collaboration plugin, menu, toolbar, and product demo rows

The two real raw Slate candidates are:

readOnly browser guard: read-only editable ignores typing and exposes the correct non-textbox DOM state.
Soft-break browser guard: Shift+Enter routes through the browser path as a distinct soft-break command.

Everything else is either already covered in live .tmp/slate-v2 tests or belongs to Plate.

Intent And Boundary

Intent: turn the TipTap harvest into only the raw Slate candidate tests.

Desired outcome: a short execution plan that tells a later implementation pass which Slate v2 tests to add, which current tests to trust, and which TipTap rows to leave out.

In scope:

raw Slate v2 browser/unit tests only
current .tmp/slate-v2 source/test grounding
de-dupe against existing Slate tests before adding anything

Non-goals:

no Plate tests
no TipTap extension API adoption
no serializer/markdown/link/toolbar/plugin work in raw Slate
no .tmp/slate-v2 implementation changes in this ralplan
no issue closure claims

Decision boundaries:

decide test ownership and target files without asking again
reject TipTap API shape when the invariant belongs to Plate or ProseMirror
keep completion pending because this is a newly created Slate Ralplan pass

Decision Brief

Principles:

Raw Slate owns primitives: input, selection, clipboard, history, render/runtime contracts.
Plate owns product policy: link range UX, markdown, serializers, menus, extension commands, collaboration plugins.
Do not port API shape; port only behavior invariants.
Prefer strengthening existing Slate tests over adding duplicate examples.

Options:

Option	Result	Verdict
Port every portable-mixed TipTap row	Imports Plate/plugin/product behavior into raw Slate and duplicates existing Slate proof.	rejected
Add no Slate tests	Misses two cheap browser guards exposed by the harvest.	rejected
Add only raw Slate gaps and mark existing coverage explicitly	Keeps Slate unopinionated and makes execution small.	chosen

Consequences:

Later execution should be tiny: probably two Playwright rows.
Most TipTap work shifts to Plate backlog, not Slate.
Issue ledgers stay unchanged unless later implementation finds an actual issue mapping.

Slate Candidate Matrix

ID	TipTap source	Slate invariant	Live Slate owner	Decision	Target
TT-SLATE-001	`../tiptap/packages/core/src/__tests__/transformPastedHTML.test.ts:8`, `:84`, `:153`, `:204`; `../tiptap/tests/cypress/integration/core/transformPastedHTML.spec.ts`	Rich HTML paste must import deterministically and avoid illegal kernel transitions. Do not port TipTap extension transform order to raw Slate.	`.tmp/slate-v2/playwright/integration/examples/paste-html.test.ts:117`, `:183`, `:942`	covered / keep	No new raw Slate test. Run paste-html browser gate when clipboard code changes.
TT-SLATE-002	`../tiptap/demos/src/GuideContent/ReadOnly/React/index.spec.js:12`, `:23`	Read-only editable must reject text input and expose non-editable DOM semantics.	Current source: `.tmp/slate-v2/packages/slate-react/src/components/editable.tsx:97`, `:220`, `:281`, `:285`, `:300`; browser owner: `.tmp/slate-v2/playwright/integration/examples/read-only.test.ts`.	strengthened	Added typing-no-op assertion to the existing read-only browser row.
TT-SLATE-003	`../tiptap/demos/src/Nodes/HardBreak/React/index.spec.js:12`, `:19`, `:34`, `:42`	`Shift+Enter` must route through browser input as the `soft` break command. Do not port TipTap `mod+Enter`; Slate's open-line contract is different.	Hotkey source/test: `.tmp/slate-v2/packages/slate-dom/src/utils/hotkeys.ts:27`, `.tmp/slate-v2/packages/slate-dom/test/hotkeys.ts:37`; runtime mapping: `.tmp/slate-v2/packages/slate-react/src/editable/editing-kernel.ts:942`; browser owner: `.tmp/slate-v2/playwright/integration/examples/richtext.test.ts`.	created	Added a focused richtext browser row for `Shift+Enter` soft command routing.
TT-SLATE-004	`../tiptap/demos/src/Extensions/UndoRedo/React/index.spec.js:19`, `:25`, `:33`, `:43`, `:69`, `:79`	Undo/redo hotkeys must include non-English physical-key fallback.	`.tmp/slate-v2/packages/slate-dom/test/hotkeys.ts:54`; history browser rows in `.tmp/slate-v2/playwright/integration/examples/plaintext.test.ts:271`, `:302` and `richtext.test.ts:4467`, `:4499`, `:4531`.	covered / keep	No new test unless Playwright can reliably emit Cyrillic key plus physical code in a real browser row.
TT-SLATE-005	`../tiptap/packages/core/__tests__/unmounted.spec.ts:8`, `:29`, `:69`, `:128`, `:148`, `:166`, `:184`, `:202`; `../tiptap/packages/core/__tests__/dispatchTransaction.spec.ts:8`, `:26`, `:56`, `:75`, `:102`	Mount/unmount/remount must not duplicate input handling; raw Slate should not adopt TipTap dispatch middleware.	`.tmp/slate-v2/playwright/integration/examples/richtext.test.ts:4614`; store destroy owners exist in `projection-store.ts` and `annotation-store.ts`.	covered / reject TipTap middleware	No new test. Keep current remount row.
TT-SLATE-006	`../tiptap/packages/core/__tests__/isNodeEmpty.spec.ts:21`, `:42`, `:74`, `:100`, `:134`, `:147`, `:192`	Empty element checks must classify empty text/block/void behavior consistently.	`.tmp/slate-v2/packages/slate/test/query-contract.ts:988`, `:1018`, `:1047`, `:2445`	covered / keep	No new test. Existing query contracts are stronger than TipTap row.
TT-SLATE-007	`../tiptap/demos/src/Examples/Performance/React/index.spec.js:6`	Large editor route should mount and be testable.	`.tmp/slate-v2/playwright/integration/examples/huge-document.test.ts:7`, `:21`; stress helpers verified by `slate-browser test:proof`.	covered / defer	No TipTap-derived perf test. Use existing huge-document and stress gates.
TT-SLATE-008	`../tiptap/demos/src/Commands/InsertContent/React/index.spec.js`; `../tiptap/demos/src/Commands/SetContent/React/index.spec.js`	Synthetic paste/text insertion imports data and updates selection.	`.tmp/slate-v2/playwright/integration/examples/plaintext.test.ts:52`; rich paste rows under `paste-html.test.ts`.	covered / keep	No new raw parser/serializer rows; Plate owns broad insert-content policy.

Execution Plan

Phase 1: add readOnly browser proof. Completed.

File: .tmp/slate-v2/playwright/integration/examples/read-only.test.ts
Route: /examples/read-only
Assertions:
- root has data-slate-editor
- root has contenteditable="false"
- root has no textbox role from Editable
- keyboard text attempt leaves model/rendered text unchanged
Gate run from .tmp/slate-v2: PLAYWRIGHT_RETRIES=0 ./node_modules/.bin/playwright test playwright/integration/examples/read-only.test.ts --project=chromium -> 1 passed.

Phase 2: add soft-break browser proof. Completed.

File: likely .tmp/slate-v2/playwright/integration/examples/richtext.test.ts
Action:
- collapse selection inside normal paragraph
- press Shift+Enter
- assert the kernel records { kind: 'insert-break', variant: 'soft' }
- leave current block-split semantics to existing core contract coverage
Explicit reject: do not add mod+Enter; TipTap treats it as hard-break, Slate maps open-line separately.
Gate run from .tmp/slate-v2: PLAYWRIGHT_RETRIES=0 ./node_modules/.bin/playwright test playwright/integration/examples/richtext.test.ts --project=chromium --grep "records a soft break command" -> 1 passed.

Phase 3: no-op coverage audit for covered rows.

Do not add tests for paste transform order, non-English undo hotkeys, lifecycle/remount, empty-node basics, or performance smoke unless the focused execution pass finds the cited live tests missing or stale.
Gates already run in this ralplan:
- cwd=/Users/zbeyens/git/slate-v2 bun test ./packages/slate-dom/test/hotkeys.ts -> 14 pass
- cwd=/Users/zbeyens/git/slate-v2 bun test ./packages/slate-history/test/history-contract.ts -> 24 pass
- cwd=/Users/zbeyens/git/slate-v2 bun test ./packages/slate/test/snapshot-contract.ts ./packages/slate/test/transaction-contract.ts -> 228 pass
- cwd=/Users/zbeyens/git/slate-v2 bun --filter slate-browser test:proof -> 23 pass
Browser gates not run in this ralplan; they belong to implementation.

Ecosystem Strategy Synthesis

System	Source	Mechanism	Avoids	Steal	Reject	Slate target	Verdict
TipTap	`transformPastedHTML` tests	Ordered extension paste transforms over ProseMirror view props.	nondeterministic plugin paste chains	Deterministic paste proof shape.	Raw `transformPastedHTML` extension chain.	Existing paste-html gauntlets; Plate paste pipeline can own extension order later.	partial
TipTap	`ReadOnly` guide spec	Browser asserts read-only edit attempt and focusability state.	readOnly prop only tested as static render output	Add Slate browser guard.	TipTap `setEditable` API.	`/examples/read-only` Playwright row.	agree
TipTap	`HardBreak` node spec	Browser/shortcut proof for soft-break command routing.	hotkey unit tests without runtime DOM proof	Add Slate browser soft-break row.	`mod+Enter` hard-break behavior and TipTap hard-break node semantics.	Richtext soft-break Playwright row.	partial
TipTap	`UndoRedo` spec	UI and non-English shortcut coverage.	English-only hotkey assumptions	Keep non-English physical-key unit coverage.	TipTap button/UI state.	Existing Slate DOM hotkey row.	agree
TipTap	`unmounted` spec	Lifecycle tests for unmounted editor.	duplicate event handlers after remount	Keep route-remount browser proof.	TipTap mount/unmount events and dispatch middleware.	Existing remount row.	diverge
TipTap	`isNodeEmpty` spec	Empty-node classification matrix.	ambiguous empty/void semantics	Keep broad query contracts.	ProseMirror hardBreak/mention node taxonomy.	Existing `Editor.isEmpty` query contracts.	partial

Issue Accounting

ClawSweeper status: skipped.

Reason: this plan is a test-harvest routing plan with no issue-fix claim, no PR narrative change, no public API change, and no implementation. If the later execution pass adds browser rows and discovers a matching open issue, run ClawSweeper then before claiming anything.

Fixed issues: none.

Improved issues: none.

Related issues: none claimed.

PR description: unchanged, because this plan changes no issue claims, API shape, release gate, example narrative, or implementation.

Applicable Review Matrix

Lens	Status	Reason	Plan delta
`tdd`	applied	The output is executable acceptance criteria for two browser rows.	Add readOnly and soft-break as tests before implementation changes.
`vercel-react-best-practices`	skipped	No React rendering implementation change in this plan.	None.
`performance-oracle`	applied lightly	TipTap performance spec is only a smoke row; live Slate has stronger 10k route and stress helpers.	Defer TipTap perf; keep existing huge-document/stress gates.
`performance`	skipped	No production perf claim or dashboard work.	None.
`build-web-apps:shadcn`	skipped	No UI component/chrome work.	None.
`react-useeffect`	skipped	No effects or external-system sync change.	None.

Confidence Scorecard

Dimension	Score	Evidence
React 19.2 runtime performance	0.88	No render implementation change; performance row mapped to existing huge-document and stress proof files.
Slate-close unopinionated DX	0.94	Rejects TipTap command-chain and extension APIs; only raw browser primitives stay.
Plate and slate-yjs migration backbone	0.88	Plate-owned rows excluded; collaboration rows remain Plate/Yjs backlog, not raw Slate.
Regression-proof testing strategy	0.95	Read-only and soft-break browser rows now exist and passed focused Playwright gates.
Research evidence completeness	0.90	TipTap harvest plus live `.tmp/slate-v2` source/tests cited; no external docs needed.
shadcn-style composability and hook/component minimalism	0.91	No new props/hooks; tests exercise existing surfaces.
Weighted total	0.95	Execution slice implemented both candidate browser rows and passed focused gates.

Pass-State Ledger

Pass	Status	Evidence added	Plan delta	Open issues	Next owner
Current-state read and initial score	complete	TipTap harvest files, live Slate source/tests, focused unit/proof commands.	Raw Slate candidate matrix created.	Browser proof rows not implemented.	user review / `ralph` execution prompt
Ralph execution	complete	`read-only.test.ts` strengthened; `richtext.test.ts` soft command row added; focused Playwright gates passed.	TT-SLATE-002 and TT-SLATE-003 moved to implemented.	None.	verification closeout
Closure score and final gates	complete	Focused Playwright rows, hotkeys gate, slate-browser proof, lint, root typecheck.	Plan status moved to `done`.	None.	complete

Ralph Execution Evidence

Changed files in .tmp/slate-v2:

playwright/integration/examples/read-only.test.ts
playwright/integration/examples/richtext.test.ts

Commands run from .tmp/slate-v2:

PLAYWRIGHT_RETRIES=0 ./node_modules/.bin/playwright test playwright/integration/examples/read-only.test.ts --project=chromium -> 1 passed
PLAYWRIGHT_RETRIES=0 ./node_modules/.bin/playwright test playwright/integration/examples/richtext.test.ts --project=chromium --grep "records a soft break command" -> 1 passed
bun test ./packages/slate-dom/test/hotkeys.ts -> 14 passed
bun --filter slate-browser test:proof -> 23 passed
bun lint:fix -> checked 1601 files, no fixes applied
bun typecheck:root -> passed

Rejected tactic:

Do not add a duplicate read-only test file; the live browser owner already existed and only needed the TipTap typing-no-op assertion.
Do not port TipTap mod+Enter hard-break behavior.
Do not force TipTap hard-break node semantics into raw Slate.

Final Completion Gates

Before this lane can be called complete after execution:

readOnly browser row exists and passes from .tmp/slate-v2
soft-break browser row exists and passes from .tmp/slate-v2
covered rows are not duplicated
no Plate-owned TipTap rows are added to raw Slate
bun test ./packages/slate-dom/test/hotkeys.ts still passes
relevant focused Playwright rows pass
completion state can move from pending to done only after the closure pass