Slate v2 PortableText Harvest Plan

Objective: Produce a lane-specific editor-harvest-plan for slate-v2 from docs/editor-test-harvester/portabletext/report.md. Completion means this dated plan records the Portable Text license mode, inventory/test-index/source-routing status, accounts for every harvest family and every source-routing row as in-lane, split, out-of-lane, skip, or duplicate, maps current .tmp/slate-v2 owner coverage for every raw Slate family, applies the downstream slate-plan gates, writes an execution queue and accepted-plan handoff, records issue/claim accounting as unchanged, edits no implementation/test/example/package/build files, and passes node .agents/rules/autogoal/scripts/check-complete.mjs docs/plans/2026-05-29-slate-v2-portabletext-harvest-plan.md.

Goal plan: docs/plans/2026-05-29-slate-v2-portabletext-harvest-plan.md

Template: docs/plans/templates/editor-harvest-plan.md

Applied packs:

editor-harvest-plan
slate-plan downstream lane gates

Completion threshold:

Score is 0.95, with every dimension at or above 0.92.
Harvest report exists at docs/editor-test-harvester/portabletext/report.md and has status: done.
License mode is permissive, output mode is durable, and the plan uses fresh invariant wording plus source provenance only.
Inventory/test-index/source-routing artifacts exist and are accounted: 502 inventory rows, 321 runnable rows, 200 portable or portable-mixed indexed files, 1944 anchors, and 200 source-routing rows.
All 14 Portable Text harvest families are routed: 7 in-lane raw Slate families, 5 split raw/product families, 2 out-of-lane Plate families, plus harness/skip inventory buckets.
No in-lane family lacks owner coverage, action, target, proof route, and verification command or explicit defer reason.
Downstream slate-plan gates are applied as planning-only boundaries.
Accepted-plan execution handoff is complete.
No .tmp/slate-v2 implementation, test, example, package, or build file is edited in this planning pass.
The autogoal mechanical completion command passes for this exact plan path.

Verification surface:

Harvest artifacts: docs/editor-test-harvester/portabletext/report.md, inventory.md, test-index.md, and source-routing.md.
Harvest report facts: status: done, score 0.94, license mode permissive, output mode durable, 502 inventory rows, 200 portable/portable-mixed source-routing rows, 100 Plate-owned rows, 188 harness rows, 14 skip rows, and 0 uncertain rows.
Current owner coverage search: rg --files and rg -n over .tmp/slate-v2/packages/slate/test, .tmp/slate-v2/packages/slate-react/test, .tmp/slate-v2/packages/slate-browser/test, .tmp/slate-v2/packages/slate-dom/test, and .tmp/slate-v2/playwright.
Downstream lane skill: .agents/skills/slate-plan/SKILL.md.
Plan integrity grep: required sections for harvest grounding, lane contract, full row accounting, in-lane candidate matrix, execution queue, downstream lane application, and accepted-plan execution handoff.
Final mechanical gate: node .agents/rules/autogoal/scripts/check-complete.mjs docs/plans/2026-05-29-slate-v2-portabletext-harvest-plan.md.

Constraints:

This is planning/routing work only.
Do not edit .tmp/slate-v2 from this skill.
Do not patch Plate packages, docs, examples, registry, package files, or build config from this skill.
Do not import Portable Text schema, behavior API names, list semantics, renderer policy, serializer outputs, toolbar/plugin policy, or Sanity bridge shape into raw Slate.
Browser, clipboard, selection, drag/drop, focus, and IME rows need honest browser/device proof when executed later.
The live .tmp/slate-v2 checkout wins over stale plans.

Boundaries:

Requested lane: slate-v2.
Harvest report: docs/editor-test-harvester/portabletext/report.md.
Source routing appendix: docs/editor-test-harvester/portabletext/source-routing.md.
Raw Slate owner: .tmp/slate-v2/packages/slate, .tmp/slate-v2/packages/slate-react, .tmp/slate-v2/packages/slate-browser, .tmp/slate-v2/packages/slate-dom, and .tmp/slate-v2/playwright.
Plate owner: product schema, lists, annotations, decorators, serializers, renderers, toolbars, plugin ergonomics, examples, and Sanity bridge work in this checkout.
Allowed edit scope in this pass: this plan file only.
Non-goals: implementation, tests, examples, package changes, GitHub issue changes, commits, pushes, PRs, or browser proof execution.

Blocked condition: No blocker is active. A real blocker would be missing docs/editor-test-harvester/portabletext/report.md, missing .tmp/slate-v2, or a user decision to route schema/list/plugin rows into raw Slate despite the lane boundary.

Lane state:

harvest_report: done at docs/editor-test-harvester/portabletext/report.md
lane: slate-v2
current_pass: execution-grade routing plan
current_pass_status: complete
current_pass_skill: .agents/skills/editor-harvest-plan/SKILL.md
downstream_skill: .agents/skills/slate-plan/SKILL.md
next_pass: user review, then downstream execution if accepted
goal_status: complete after final mechanical gate succeeds

Current verdict:

verdict: accept plan
score: 0.95
next owner: downstream Slate v2 execution pass, likely slate-plan or slate-patch based on whether the user wants architecture-first implementation or bug-style patch execution.
reason: Portable Text is valuable as behavior pressure for raw selection, projection, void/object boundaries, insert/delete, clipboard, drag/drop, IME, focus/event authority, collab/history, operations, and transaction ordering; its schema/product API shape belongs outside raw Slate.

Completion rule:

Call update_goal(status: complete) only after this file has checked Work Checklist items, resolved Start Gates and Completion Gates, phase rows with closed statuses, Verification evidence, Reboot status, Open risks, and a passing autogoal check.
Do not start implementation from this skill.

Start Gates:

Gate	Applies	Evidence
Skill analysis before edits	yes	`editor-harvest-plan` and downstream `slate-plan` skill bodies were read before this plan was written.
Active goal checked or created	yes	Active goal `019e6fd0-9a50-7120-ac35-068d11f61bcc` matches this exact plan path and completion command.
Harvest source read	yes	`docs/editor-test-harvester/portabletext/report.md` and `source-routing.md` were read; companion artifacts were checked for presence.
Current owner evidence searched	yes	`.tmp/slate-v2` package tests, browser tests, DOM tests, and Playwright examples were searched before coverage mapping.

Work Checklist:

Completion Gates:

Gate	Applies	Required action	Evidence
Named verification threshold	yes	Verify harvest artifacts, row accounting, downstream gates, and final autogoal check.	Verification evidence records artifact checks, section greps, placeholder scan, and autogoal command.
Harvest artifacts current	yes	Confirm report, inventory, test-index, and source-routing exist and match the 2026-05-29 harvest.	`harvest-files-ok` returned for all companion files; report records 502 inventory rows and 200 routed portable/mixed files.
Row accounting closed	yes	Account for all 14 families and all 200 source-routing rows.	Family table accounts 94 in-lane routed files, 106 split routed files, 100 out-of-lane Plate-owned inventory files, and 202 harness/skip inventory files.
Current owner mapping searched	yes	Search current `.tmp/slate-v2` coverage before target decisions.	Owner coverage section lists live test/browser anchors found by `rg`.
Downstream lane gates applied	yes	Apply `slate-plan` raw Slate/Plate boundaries and planning-only rule.	Downstream lane application table records current-source, raw Slate owner, Plate split, browser proof, and no-implementation boundaries.
Issue or claim accounting changed	no	Record that no issue, PR, or claim ledger changed.	Issue and claim accounting says no fixed/improved/related claim changes.
Accepted-plan handoff	yes	Emit execution queue, broad final gate, issue sync rule, and stop rule.	Accepted-plan execution handoff section names queue IDs and commands.
Goal plan complete	yes	Run the autogoal completion checker for this exact plan path.	Final command result is recorded in Verification evidence.

Phase / pass table:

Phase	Status	Evidence	Next
Harvest grounding	complete	Report, inventory, test-index, and source-routing facts recorded.	done
Lane filter and row accounting	complete	14 families, 200 source-routing rows, and 502 inventory rows accounted.	done
Current owner coverage mapping	complete	`.tmp/slate-v2` package and browser owner files searched and mapped.	done
Execution queue and proof routing	complete	Queue `PT-SV2-Q1` through `PT-SV2-Q7` written with commands and boundaries.	done
Downstream lane gate application	complete	`slate-plan` raw/Plate boundaries recorded.	done
Issue and claim accounting	complete	No issue, PR, or claim status changed.	done
Accepted-plan execution handoff	complete	Handoff section names plan path, lane, queue, commands, broad gate, and stop rule.	done
Closure review and final gates	complete	Mechanical checks recorded below.	final response

Confidence score:

Dimension	Weight	Score	Evidence	Cap hit
Harvest source readiness	0.15	0.98	MIT license mode, durable report, 502 inventory rows, 321 runnable rows, 200 indexed portable/mixed files, 1944 anchors.	none
Lane-filter completeness	0.25	0.96	All 14 harvest families and all 200 source-routing rows are classified; 0 uncertain rows remain.	none
Current owner coverage mapping	0.25	0.94	Live `.tmp/slate-v2` owner searches found coverage anchors for each raw family.	no behavior tests run because this is planning-only
Actionability of execution queue	0.20	0.94	Each queue row has ID, target owner, proof kind, focused command, split rule, and broad gate.	browser rows still need execution in a later pass
License/provenance discipline	0.15	0.96	Plan records permissive license mode and uses behavior invariants plus source refs only.	none
Weighted total	1.00	0.95	Passes the 0.92 threshold and no dimension is below 0.85.	none

Harvest grounding:

Field	Value
Harvest report	`docs/editor-test-harvester/portabletext/report.md`
Status	done
Harvest score	0.94
Source target	`../portabletext`
License mode	permissive; MIT evidence recorded in `../portabletext/LICENSE` by the harvest report
Output mode	durable
Inventory status	`inventory.md` exists; 502 classified rows; 321 runnable rows; 94 portable; 106 portable-mixed; 100 Plate-owned; 188 harness; 14 skip; 0 product-shell; 0 uncertain
Test-index status	`test-index.md` exists; 200 portable/portable-mixed indexed runnable files; 1944 anchors
Source-routing status	`source-routing.md` exists; 200 portable/portable-mixed files routed
Matrix families	14 total: `PT-H01` through `PT-H12`, `PT-P01`, `PT-P02`
Inventory equation	94 in-lane portable + 106 split portable-mixed + 100 Plate-owned + 188 harness + 14 skip = 502

Lane contract:

Field	Value
Lane	`slate-v2`
Aliases	`slate`, `raw-slate`, `Slate v2 substrate`
Downstream skill	`.agents/skills/slate-plan/SKILL.md`
Owner boundary	Raw Slate owns document model operations, path/point/range behavior, root projection, DOM selection mapping, native input transport, clipboard fragments, drag/drop validity, void/object atomics, history/collab rebasing, and transaction ordering.
Exclusions	Plate owns schema/category admissibility, lists, annotations, decorators, comments, serializers, renderer definitions, toolbar/plugin ergonomics, Sanity bridge behavior, and docs/examples for product-level features.
Planning rule	This pass routes and plans only; execution starts only after user acceptance through the downstream lane skill.

Full harvest row accounting: Every row in docs/editor-test-harvester/portabletext/source-routing.md is accounted through its family bucket below. The row count for source-routing is 200. PT-H10 is a matrix-only focus/event authority family with 0 direct source-routing files, so it is routed for raw API pressure without changing the 200-row count.

Family	Family title	Files	Anchors	Harvest category	Lane accounting	Slate v2 decision	Source-routing status
PT-H01	Selection boundary and projection	24	178	portable	in-lane	refactor-existing plus create exact root/object endpoint gaps	24 of 200 routed
PT-H02	Content roots and containers	30	241	portable-mixed	split	raw root projection and navigation in Slate; schema envelope to Plate	30 of 200 routed
PT-H03	Void and object boundary editing	8	75	portable	in-lane	refactor-existing plus browser rows for object-boundary gaps	8 of 200 routed
PT-H04	Insert, split, and break placement	17	236	portable	in-lane	refactor-existing placement matrix	17 of 200 routed
PT-H05	Delete and backspace matrix	8	84	portable	in-lane	refactor-existing plus object/root boundary gaps	8 of 200 routed
PT-H06	Clipboard and fragment serialization	3	15	portable-mixed	split	raw fragment/selection proof in Slate; serializer and matchers to Plate	3 of 200 routed
PT-H07	Drag/drop and drop target resolution	4	33	portable-mixed	split	native selection/drop validity in Slate; handles/product DnD to Plate	4 of 200 routed
PT-H08	History and remote collaboration rebasing	6	54	portable	in-lane	refactor-existing plus remote selection dedupe gaps	6 of 200 routed
PT-H09	IME and composition transport	1	25	portable	in-lane	refactor-existing; add only exact boundary transport gaps	1 of 200 routed
PT-H10	Focus, keyboard, and event authority	0	0	matrix-only raw pressure	split	raw event authority in Slate; Portable Text behavior names rejected	0 direct rows; covered by owner search and H12 pressure
PT-H11	Path, point, range, and operation core	30	374	portable	in-lane	covered/refactor-existing unless a root-aware variant is missing	30 of 200 routed
PT-H12	Behavior pipeline and extension ordering	69	629	portable-mixed	split	raw transaction/effect ordering only if accepted; plugin ergonomics to Plate	69 of 200 routed
PT-P01	Schema, renderers, annotations, decorators, and lists	29	187	plate-owned	out-of-lane	no raw Slate execution	inventory only; not in source-routing
PT-P02	Serializers, toolbar, plugins, and product packages	71	1007	plate-owned	out-of-lane	no raw Slate execution	inventory only; not in source-routing
Harness bucket	Fixtures, helpers, generated setup	188	n/a	harness	skip	no editor-behavior execution owner	inventory only
Skip bucket	Non-editor/package rows	14	n/a	skip	skip	no editor-behavior execution owner	inventory only

Row totals:

Source-routing rows: 200 = 24 + 30 + 8 + 17 + 8 + 3 + 4 + 6 + 1 + 0 + 30 + 69.
In-lane routed rows: 94 = PT-H01 + PT-H03 + PT-H04 + PT-H05 + PT-H08 + PT-H09 + PT-H11.
Split routed rows: 106 = PT-H02 + PT-H06 + PT-H07 + PT-H12; PT-H10 has 0 direct source files but remains split API pressure.
Out-of-lane inventory rows: 100 = PT-P01 + PT-P02.
Skipped inventory rows: 202 = 188 harness + 14 skip.
Duplicate rows: 0; no family is used as a duplicate sink.
Unresolved rows: 0.

Current Slate v2 owner coverage:

Family	Live owner coverage found	Coverage verdict
PT-H01	`selection-rebase-contract.ts`, `projections-and-selection-contract.test.tsx`, `view-selection-contract.test.ts`, `selection-reconciler-contract.test.tsx`, browser selection tests, multi-root Playwright example	strong, add exact root/object endpoint rows only
PT-H02	`content-root-navigation-contract.test.ts`, `root-interaction-resolver.test.ts`, `runtime-live-state-contract.test.ts`, `multi-root-document.test.ts`, projected clipboard tests	strong for navigation/projection, split schema policy out
PT-H03	`slate-void-shell-contract.test.tsx`, `public-element-void-kind-contract.ts`, void transform fixtures, `editable-voids.test.ts`, generated stress rows	strong, object-boundary native selection still worth focused browser rows
PT-H04	`transforms-contract.ts`, insertNodes/insertText/insertFragment/splitNodes fixture directories, operation contracts	broad fixture coverage, refactor into matrix if exact Portable Text pressure is scattered
PT-H05	delete transform fixtures, `operations-contract.ts`, void delete fixtures, generated stress rows	broad fixture coverage, fill root/object edge cases after dedupe
PT-H06	`clipboard-contract.ts`, `clipboard-boundary.test.ts`, `projected-clipboard-contract.test.ts`, `paste-html.test.ts`	raw fragment coverage exists, browser serializer policy split remains
PT-H07	`root-interaction-resolver.test.ts`, root interaction tests, browser selection examples, generated editing stress utilities	unit coverage exists; exact drag/drop browser rows look weaker
PT-H08	`collab-selection-stress-contract.ts`, `collab-document-state-contract.ts`, `projected-collab-substrate-contract.test.ts`, `collab-history-runtime-contract.ts`	strong, add exact dedupe/rebase rows only
PT-H09	`playwright-ime.test.ts`, `composition-state-contract.test.ts`, `keyboard-input-strategy-contract.test.ts`, generated stress rows	strong enough to refactor-existing first
PT-H10	`selection-controller-contract.test.ts`, `selection-side-effect-policy-contract.test.ts`, `focus-slate-editable-contract.test.ts`, `hotkeys.test.ts`, `keyboard-input-strategy-contract.test.ts`	raw authority coverage exists; reject Portable Text behavior naming
PT-H11	`root-location-contract.ts`, `rooted-operation-contract.ts`, `generic-operation-contract.ts`, `operations-contract.ts`, `range-ref-contract.ts`, path/point/range fixture directories	strong, add only missing root-aware variants
PT-H12	`transaction-contract.ts`, `transaction-target-runtime-contract.ts`, root operation middleware tests, runtime provider contracts	API pressure only; do not rename around Portable Text behavior API

In-lane candidate matrix:

Queue	Families	Lane reason	Action	Target owner	Proof route
PT-SV2-Q1	PT-H01, PT-H02	Selection projection and content-root ordering are raw Slate substrate.	refactor-existing plus create exact root/object endpoint and nested-root fragment gaps	`packages/slate/test/selection-rebase-contract.ts`, `packages/slate-react/test/projections-and-selection-contract.test.tsx`, `view-selection-contract.test.ts`, `content-root-navigation-contract.test.ts`, `playwright/integration/examples/multi-root-document.test.ts`	focused package tests; Playwright only when DOM projection changes
PT-SV2-Q2	PT-H03, PT-H04, PT-H05	Void/object boundaries plus insert/delete placement are raw transforms and native selection runtime.	refactor-existing plus object-boundary browser rows for missing native behavior	`slate-void-shell-contract.test.tsx`, Slate transform fixtures, `transforms-contract.ts`, `operations-contract.ts`, `editable-voids.test.ts`, generated stress rows	focused Slate package tests plus browser proof for native caret/selection
PT-SV2-Q3	PT-H06, PT-H07	Clipboard fragments and drag/drop target validity cross raw DOM selection and projected path law.	split; create exact browser rows only after resolver/unit gaps are known	`projected-clipboard-contract.test.ts`, `clipboard-boundary.test.ts`, `paste-html.test.ts`, `root-interaction-resolver.test.ts`, browser example/stress rows	package tests plus Playwright clipboard/drag proof
PT-SV2-Q4	PT-H08	Remote selection and history rebasing are raw collaboration substrate.	refactor-existing plus exact selection emission dedupe rows if missing	`collab-selection-stress-contract.ts`, `collab-history-runtime-contract.ts`, `collab-document-state-contract.ts`, `projected-collab-substrate-contract.test.ts`	package tests
PT-SV2-Q5	PT-H09, PT-H10	IME/focus/keyboard authority belongs to browser input transport and selection authority.	refactor-existing; add exact boundary transport rows only	`playwright-ime.test.ts`, `composition-state-contract.test.ts`, `selection-controller-contract.test.ts`, `selection-side-effect-policy-contract.test.ts`, `hotkeys.test.ts`	slate-browser and slate-react focused tests; browser proof for native composition
PT-SV2-Q6	PT-H11, PT-H12	Root-aware path/operation semantics are raw; behavior pipeline only matters when reduced to transaction ordering/effects.	refactor-existing for operations; defer public API reshaping until a `slate-plan` API decision accepts it	`root-location-contract.ts`, `rooted-operation-contract.ts`, `generic-operation-contract.ts`, `operations-contract.ts`, `range-ref-contract.ts`, `transaction-contract.ts`	package tests; no Plate plugin API claim
PT-SV2-Q7	PT-P01, PT-P02, harness, skip	Product rows and harness rows are not raw Slate execution work.	out-of-lane or skip	Plate packages/docs or no owner	no `.tmp/slate-v2` command

Split rows:

Family	Raw Slate-owned part	Out-of-lane part	Decision
PT-H02	Root ordering, projected selection, normalization boundaries, DOM projection.	Schema category, allowed child policy, Portable Text path/key serialization.	Split and execute only raw root behavior in `.tmp/slate-v2`.
PT-H06	Clipboard fragment shape, object boundary projection, undo grouping over raw fragments.	HTML/markdown matchers, serializer output, MIME/product upload policy.	Split and keep serializers in Plate.
PT-H07	Native selection/drop validity, self-drop suppression, projected drop target path.	Drag handles, product DnD UX, renderer data attributes.	Split and require browser proof for raw rows.
PT-H10	Focus import, keyboard default-action authority, selection side-effect policy.	Portable Text behavior event names and plugin authoring model.	Split; keep raw event authority, reject API names.
PT-H12	Transaction ordering, commit/effect boundaries if accepted by Slate API.	Plugin behavior ergonomics, action names, renderer/list/decorator pipeline.	Treat as architecture pressure, not implementation-by-copy.

Excluded or skipped rows:

Bucket	Count	Reason	Owner
PT-P01	29	Schema, renderer definitions, annotations, decorators, comments, and lists are product/editor policy.	Plate
PT-P02	71	Serializers, toolbar, plugins, media, Sanity bridge, and product packages are not raw Slate substrate.	Plate
Harness	188	Fixtures, helper tests, generated setup, and test scaffolding do not create portable editor behavior rows.	no execution owner
Skip	14	Non-editor/package rows do not pressure Slate v2.	no execution owner

Coverage dedupe:

Candidate	Existing coverage	Decision	Evidence
IME base composition	`playwright-ime.test.ts`, `composition-state-contract.test.ts`, generated stress rows	refactor-existing first	Current owner search found direct IME/composition owners.
Basic insert/delete transforms	transform fixtures plus `transforms-contract.ts` and `operations-contract.ts`	refactor-existing before creating new files	Current owner search found broad insert/delete fixture directories.
Multi-root projection	`multi-root-document.test.ts`, `content-root-navigation-contract.test.ts`, projection tests	create only nested fragment/root gaps	Browser and package owners already exist.
Clipboard HTML paste	`projected-clipboard-contract.test.ts`, `clipboard-boundary.test.ts`, `paste-html.test.ts`	split raw fragment from Plate matcher policy	Current owner search found package and browser surfaces.
Collab history and selection	collab selection/document/history contracts	add only exact dedupe/rebase gaps	Existing collab substrate tests are live.
Transaction/behavior pipeline	`transaction-contract.ts`, root operation middleware tests	architecture decision before API surface changes	Portable Text behavior API names are not a raw Slate API model.

Execution queue:

ID	Action	Target	Proof kind	Focused verification	Notes
PT-SV2-Q1	Apply selection/content-root projection pressure from `PT-H01` and raw portions of `PT-H02`.	`.tmp/slate-v2/packages/slate/test/selection-rebase-contract.ts`; `.tmp/slate-v2/packages/slate-react/test/projections-and-selection-contract.test.tsx`; `view-selection-contract.test.ts`; `content-root-navigation-contract.test.ts`; `playwright/integration/examples/multi-root-document.test.ts`	unit plus browser if DOM projection changes	`cd .tmp/slate-v2 && bun test ./packages/slate/test/selection-rebase-contract.ts ./packages/slate-react/test/projections-and-selection-contract.test.tsx ./packages/slate-react/test/view-selection-contract.test.ts ./packages/slate-react/test/content-root-navigation-contract.test.ts`	First slice; highest value after recent root/selection bugs.
PT-SV2-Q2	Apply void/object insert/delete pressure from `PT-H03`, `PT-H04`, and `PT-H05`.	Slate transform fixtures; `transforms-contract.ts`; `operations-contract.ts`; `slate-void-shell-contract.test.tsx`; `editable-voids.test.ts`; generated stress rows	unit plus browser for native caret/selection	`cd .tmp/slate-v2 && bun test ./packages/slate/test/transforms-contract.ts ./packages/slate/test/operations-contract.ts ./packages/slate-react/test/slate-void-shell-contract.test.tsx`	Do not let hidden spacer/zero-width nodes become behavior authority.
PT-SV2-Q3	Apply clipboard/drag/drop raw portions from `PT-H06` and `PT-H07`.	`projected-clipboard-contract.test.ts`; `clipboard-boundary.test.ts`; `paste-html.test.ts`; `root-interaction-resolver.test.ts`; browser examples/stress	unit plus Playwright browser proof	`cd .tmp/slate-v2 && bun test ./packages/slate-react/test/projected-clipboard-contract.test.ts ./packages/slate-dom/test/clipboard-boundary.test.ts ./packages/slate-react/test/root-interaction-resolver.test.ts`	Browser proof required before claiming drag/drop closure.
PT-SV2-Q4	Apply remote selection/history rows from `PT-H08`.	`collab-selection-stress-contract.ts`; `collab-document-state-contract.ts`; `collab-history-runtime-contract.ts`; `projected-collab-substrate-contract.test.ts`	package unit	`cd .tmp/slate-v2 && bun test ./packages/slate/test/collab-selection-stress-contract.ts ./packages/slate/test/collab-document-state-contract.ts ./packages/slate/test/collab-history-runtime-contract.ts ./packages/slate-react/test/projected-collab-substrate-contract.test.ts`	No provider-specific Yjs product policy in raw Slate.
PT-SV2-Q5	Apply IME/focus/keyboard pressure from `PT-H09` and `PT-H10`.	`playwright-ime.test.ts`; `composition-state-contract.test.ts`; `selection-controller-contract.test.ts`; `selection-side-effect-policy-contract.test.ts`; `focus-slate-editable-contract.test.ts`; `hotkeys.test.ts`	browser input plus unit	`cd .tmp/slate-v2 && bun test ./packages/slate-browser/test/core/playwright-ime.test.ts ./packages/slate-react/test/composition-state-contract.test.ts ./packages/slate-react/test/selection-controller-contract.test.ts ./packages/slate-react/test/selection-side-effect-policy-contract.test.ts ./packages/slate-dom/test/hotkeys.test.ts`	Synthetic jsdom rows do not prove mobile/IME claims.
PT-SV2-Q6	Apply operation/path/core pressure from `PT-H11` and decide raw transaction pressure from `PT-H12`.	`root-location-contract.ts`; `rooted-operation-contract.ts`; `generic-operation-contract.ts`; `operations-contract.ts`; `range-ref-contract.ts`; `transaction-contract.ts`	package unit plus API decision	`cd .tmp/slate-v2 && bun test ./packages/slate/test/root-location-contract.ts ./packages/slate/test/rooted-operation-contract.ts ./packages/slate/test/generic-operation-contract.ts ./packages/slate/test/operations-contract.ts ./packages/slate/test/range-ref-contract.ts ./packages/slate/test/transaction-contract.ts`	Transaction/effect API changes need an accepted `slate-plan` decision first.
PT-SV2-Q7	Route Plate-owned and skipped rows out of Slate v2 execution.	Plate packages/docs or no owner	no raw Slate proof	no `.tmp/slate-v2` command	`PT-P01`, `PT-P02`, harness, and skip rows do not block raw Slate execution.

Issue and claim accounting:

Fixed issues: none from this plan.
Improved issues: none from this plan.
Related issues: unchanged; this plan routes harvested Portable Text test pressure only.
PR reference: unchanged; no claim or proof status changed.
Issue ledger sync: no ledger update applies. A later execution pass must sync claim ledgers only if it changes a public issue/claim status with exact proof.

Downstream lane application:

Gate	Status	Evidence
Downstream skill read	complete	`.agents/skills/slate-plan/SKILL.md` was read before routing decisions were finalized.
Current source wins	complete	Live `.tmp/slate-v2` owner searches were used rather than stale plan claims.
Raw Slate stays unopinionated	complete	Schema/list/plugin/serializer/toolbar/Sanity rows are excluded or split to Plate.
Browser proof honesty	complete	Clipboard, drag/drop, DOM projection, native selection, and IME rows carry browser proof routes where package tests are insufficient.
Planning-only boundary	complete	This plan records no implementation/test/example/package/build edits.
Breaking API caution	complete	`PT-H12` transaction/effect pressure requires an accepted Slate API decision before execution.

Accepted-plan execution handoff:

read-first plan path: docs/plans/2026-05-29-slate-v2-portabletext-harvest-plan.md
requested lane: slate-v2
exact execution queue IDs: PT-SV2-Q1, PT-SV2-Q2, PT-SV2-Q3, PT-SV2-Q4, PT-SV2-Q5, PT-SV2-Q6, with PT-SV2-Q7 excluded from raw Slate implementation.
implementation boundaries: a later accepted execution pass may edit .tmp/slate-v2; this planning pass edits only this plan. Keep Portable Text schema/list/plugin/product policy out of raw Slate.
focused verification commands: use the command in each execution queue row after touching that row's owners.
broad final gate: cd .tmp/slate-v2 && bun check.
browser/release-quality gate: use cd .tmp/slate-v2 && bun check:full only when browser architecture or release-quality browser claims change.
issue/claim sync rule: sync issue/claim ledgers only when the later execution pass changes a claim with proof.
stop rule: stop and reroute if a row requires schema/category/list/plugin/serializer/product semantics instead of raw editor substrate behavior.

Execution progress:

Queue	Status	Evidence	Next owner
PT-SV2-Q1	complete	Added a focused Slate React regression proving nested content roots inside another content root are ordered as visible sibling blocks when projected selection enters and exits the nested root.	done
PT-SV2-Q2	complete	Added a focused Slate transform regression proving `insertText` over a selection spanning a block void removes the void boundary and leaves one valid collapsed text selection.	done
PT-SV2-Q3	complete	Fixed projected clipboard to serialize visible repeated-root owner copies while keeping repeated-root mutation commands rejected; verified clipboard package tests and Chromium copy/drag/drop browser rows.	done
PT-SV2-Q4	complete	Discharged as already covered by live collab/history contracts for remote selection rebasing, remote state patch replay, history skip metadata, projected remote-paint policy, and runtime ID rebasing.	done
PT-SV2-Q5	complete	Discharged as already covered by live IME/composition, selection import/export authority, focus side-effect, hotkey, and keyboard-input strategy contracts, with focused Chromium IME proof.	done
PT-SV2-Q6	complete	Discharged as already covered by live root-location, rooted-operation, operation replay/inversion, range-ref, transaction, command middleware, and afterCommit contracts.	done

Execution evidence:

2026-05-29 PT-SV2-Q1: edited .tmp/slate-v2/packages/slate-react/test/content-root-navigation-contract.test.ts.
Test added: orders nested content roots inside content roots as visible sibling blocks.
Focused Slate package verification passed: cd .tmp/slate-v2 && bun test ./packages/slate/test/selection-rebase-contract.ts -> 5 pass, 0 fail.
Focused Slate React verification passed: cd .tmp/slate-v2 && bun --filter slate-react test:vitest -- test/view-selection-contract.test.ts test/content-root-navigation-contract.test.ts test/projections-and-selection-contract.test.tsx -> 3 files passed, 42 tests passed.
Slate React typecheck passed: cd .tmp/slate-v2 && bun --filter slate-react typecheck.
Browser proof not run for Q1 because this slice added package-level projection coverage only and did not change DOM projection runtime code.
Issue/reference sync unchanged: Q1 adds harvested coverage only and makes no public issue claim.
Autoreview not required for Q1: the slice is a focused test-only coverage addition with no implementation diff.
2026-05-29 PT-SV2-Q2: edited .tmp/slate-v2/packages/slate/test/transforms-contract.ts.
Test added: insertText replaces a selection spanning a block void without keeping a void anchor.
Focused Slate transform/operation verification passed: cd .tmp/slate-v2 && bun test ./packages/slate/test/transforms-contract.ts ./packages/slate/test/operations-contract.ts -> 43 pass, 0 fail.
Focused Slate React void-shell verification passed: cd .tmp/slate-v2 && bun --filter slate-react test:vitest -- test/slate-void-shell-contract.test.tsx -> 1 file passed, 1 test passed.
Slate package typecheck passed: cd .tmp/slate-v2 && bun --filter slate typecheck.
Browser proof not run for Q2 because this slice added package-level transform coverage only and did not change DOM/native void runtime code.
Issue/reference sync unchanged: Q2 adds harvested coverage only and makes no public issue claim.
Autoreview not required for Q2: the slice is a focused test-only coverage addition with no implementation diff.
2026-05-29 PT-SV2-Q3: edited .tmp/slate-v2/packages/slate-react/src/editable/projected-clipboard.ts and .tmp/slate-v2/packages/slate-react/test/projected-clipboard-contract.test.ts.
Test added: serializes repeated content-root owners as visible clipboard fragments.
Behavior fixed: projected clipboard resolves fragment ranges directly from visible selection segments, so repeated owner copies of the same child root can be copied in visible order; mutation commands still use createProjectedSelectionTarget and keep repeated-root ambiguity rejected.
Focused projected clipboard/root interaction verification passed: cd .tmp/slate-v2 && bun --filter slate-react test:vitest -- test/projected-clipboard-contract.test.ts test/root-interaction-resolver.test.ts -> 2 files passed, 13 tests passed.
Focused Slate DOM/clipboard verification passed: cd .tmp/slate-v2 && bun test ./packages/slate-dom/test/clipboard-boundary.test.ts ./packages/slate/test/clipboard-contract.ts -> 29 pass, 0 fail.
Slate React typecheck passed: cd .tmp/slate-v2 && bun --filter slate-react typecheck.
Chromium projected clipboard proof passed: cd .tmp/slate-v2 && PLAYWRIGHT_RETRIES=0 bunx playwright test playwright/integration/examples/synced-blocks.test.ts --project=chromium --grep "copies a projected selection from visible order" -> 1 passed.
Chromium drag/drop proof passed: cd .tmp/slate-v2 && PLAYWRIGHT_RETRIES=0 bunx playwright test playwright/integration/examples/synced-blocks.test.ts playwright/integration/examples/editable-voids.test.ts --project=chromium --grep "mouse drag can select from p1 through repeated synced roots to p2|drops HTML payload inside same-runtime child root without stealing outer selection" -> 2 passed.
Issue/reference sync unchanged: Q3 changes harvested behavior coverage and clipboard substrate only, with no public issue claim.
Autoreview initially found two accepted P2 root chrome coordinate defects in root-interaction-controller.ts: nested editor strings could be hit-tested against the wrong root, and RTL line-edge clicks mapped physical right to logical end.
Autoreview fixes applied: extracted coordinate placement into slate-string-coordinate-placement.ts, scoped root chrome strings to the owning editable root, mapped physical line edges through computed text direction, and added focused coverage in slate-string-coordinate-placement.test.ts.
Autoreview-fix verification passed: cd .tmp/slate-v2 && bun --filter slate-react test:vitest -- test/slate-string-coordinate-placement.test.ts test/root-interaction-resolver.test.ts test/use-slate-root-chrome.test.tsx -> 3 files passed, 16 tests passed.
Autoreview-fix typecheck passed: cd .tmp/slate-v2 && bun --filter slate-react typecheck.
Affected projected clipboard/root interaction verification passed after the autoreview fixes: cd .tmp/slate-v2 && bun --filter slate-react test:vitest -- test/projected-clipboard-contract.test.ts test/root-interaction-resolver.test.ts test/slate-string-coordinate-placement.test.ts test/use-slate-root-chrome.test.tsx -> 4 files passed, 20 tests passed.
Affected Chromium copy/drag/drop proof passed after the autoreview fixes: cd .tmp/slate-v2 && PLAYWRIGHT_RETRIES=0 bunx playwright test playwright/integration/examples/synced-blocks.test.ts playwright/integration/examples/editable-voids.test.ts --project=chromium --grep "copies a projected selection from visible order|mouse drag can select from p1 through repeated synced roots to p2|drops HTML payload inside same-runtime child root without stealing outer selection" -> 3 passed.
Autoreview rerun after the root chrome fixes returned clean: cd .tmp/slate-v2 && /Users/zbeyens/git/plate-2/.agents/skills/autoreview/scripts/autoreview --mode local -> clean, no accepted/actionable findings.
2026-05-29 PT-SV2-Q4: no .tmp/slate-v2 edits; row discharged as existing coverage after reading the collab/history owners.
Existing coverage confirmed: collab-selection-stress-contract.ts covers high-QPS remote selection rebasing, same-offset contention, suffix insert stability, split/merge rebasing, and remove-node stale path resolution.
Existing coverage confirmed: collab-document-state-contract.ts covers shared state patch replay and local-only state exclusion from collab payloads.
Existing coverage confirmed: collab-history-runtime-contract.ts covers remote metadata, history skip, three-peer convergence, paste replay, remote rebase of undo/redo batches, bookmark rebasing, and runtime ID locality under remote remove/move.
Existing coverage confirmed: projected-collab-substrate-contract.test.ts covers root-keyed projected lifecycle operations and remote selection paint policy without serializing projection owners.
Focused Slate collab/history verification passed: cd .tmp/slate-v2 && bun test ./packages/slate/test/collab-selection-stress-contract.ts ./packages/slate/test/collab-document-state-contract.ts ./packages/slate/test/collab-history-runtime-contract.ts -> 16 pass, 0 fail.
Focused Slate React projected collab verification passed: cd .tmp/slate-v2 && bun --filter slate-react test:vitest -- test/projected-collab-substrate-contract.test.ts -> 1 file passed, 3 tests passed.
Issue/reference sync unchanged: Q4 makes no public issue claim and required no implementation change.
Autoreview not required for Q4: no implementation diff.
2026-05-29 PT-SV2-Q5: no .tmp/slate-v2 edits; row discharged as existing coverage after reading the IME/focus/keyboard owners.
Existing coverage confirmed: playwright-ime.test.ts covers synthetic composition transport against the resolved editor surface and one-time composition key event installation.
Existing coverage confirmed: composition-state-contract.test.ts covers read-only compositionstart/update/end authority without mutating editor text.
Existing coverage confirmed: selection-controller-contract.ts covers model-vs-DOM selection import/export policy, nested editable ownership, native selectionchange authority, changed expanded DOM selection import, and native input preference guards.
Existing coverage confirmed: selection-side-effect-policy-contract.ts covers remote collaboration metadata suppressing DOM selection, scroll, and focus side effects.
Existing coverage confirmed: focus-slate-editable-contract.test.ts covers focus without exporting a model selection over an active projected view selection.
Existing coverage confirmed: keyboard-input-strategy-contract.test.ts covers read-only keyboard authority, model-owned key commands, history focus repair across roots, hidden boundary arrow navigation, shift extension, word extension, multiple hidden ranges, and line movement over hidden ranges.
Existing coverage confirmed: hotkeys.test.ts covers semantic key matching, physical-code fallback only for non-ASCII layouts, optional modifiers, platform mod, modifier-only keys, and AltGraph exclusion.
Focused slate-browser IME helper verification passed: cd .tmp/slate-v2/packages/slate-browser && bun test test/core/playwright-ime.test.ts -> 2 pass, 0 fail.
Focused Slate React selection/keyboard verification passed: cd .tmp/slate-v2 && bun --filter slate-react test:vitest -- test/composition-state-contract.test.ts test/selection-controller-contract.test.ts test/selection-side-effect-policy-contract.test.ts test/focus-slate-editable-contract.test.ts test/keyboard-input-strategy-contract.test.ts -> 5 files passed, 44 tests passed.
Focused slate-dom hotkey verification passed: cd .tmp/slate-v2/packages/slate-dom && bun test test/hotkeys.test.ts -> 15 pass, 0 fail.
Chromium IME proof passed: cd .tmp/slate-v2 && PLAYWRIGHT_RETRIES=0 bunx playwright test playwright/integration/examples/hovering-toolbar.test.ts playwright/integration/examples/dom-coverage-boundaries.test.ts --project=chromium --grep "keeps hovering toolbar hidden during IME composition|commits IME composition while DOM coverage boundaries are hidden" -> 2 passed.
Issue/reference sync unchanged: Q5 makes no public issue claim and required no implementation change.
Autoreview not required for Q5: no implementation diff.
2026-05-29 PT-SV2-Q6: no .tmp/slate-v2 edits; row discharged as existing coverage after reading operation/path/transaction owners.
Existing coverage confirmed: root-location-contract.ts covers implicit/explicit root resolution for operations, points, ranges, and selection patches.
Existing coverage confirmed: rooted-operation-contract.ts covers root-explicit operations, root-local point/range transforms, root-local refs, sibling root split/merge isolation, explicit non-main range edits, root lifecycle operations, root validation, selection operation inversion, and view-root routing.
Existing coverage confirmed: operations-contract.ts covers replace operations, invalid replay rollback, ref rebasing, move semantics, selection rebasing over split/merge/remove operations, and protected raw node property checks.
Existing coverage confirmed: range-ref-contract.ts covers commit-time ref publication, affinity, move/split/merge rebasing, rollback survival, rootless view-root refs, explicit non-main roots, and matching-root invalidation only.
Existing coverage confirmed: transaction-contract.ts covers batch/manual parity, draft visibility, one publish per transaction, rollback, command middleware, commit metadata, command priority, extension registry cleanup, transaction groups, and commit listener delivery.
Existing coverage confirmed: update-after-commit-contract.ts covers update-local afterCommit effects, no-commit/drop-on-rollback behavior, nested update ordering, snapshot capture before later updates, onCommit ordering, view-scoped snapshots, and stale registration rejection.
Focused operation/path/transaction verification passed: cd .tmp/slate-v2 && bun test ./packages/slate/test/root-location-contract.ts ./packages/slate/test/rooted-operation-contract.ts ./packages/slate/test/generic-operation-contract.ts ./packages/slate/test/operations-contract.ts ./packages/slate/test/range-ref-contract.ts ./packages/slate/test/transaction-contract.ts ./packages/slate/test/update-after-commit-contract.ts -> 108 pass, 0 fail.
Issue/reference sync unchanged: Q6 makes no public issue claim and required no implementation change.
Autoreview not required for Q6: no implementation diff.
2026-05-29 closeout: broad bun check initially failed after lint/typecheck/Bun tests on stale React contract inventories, package peer floor drift, and a pagination example full-value replacement after mount.
Broad-gate fixes applied: aligned slate-dom and slate-react peer floors to the live sibling 0.124.1 packages, updated explicit ownership inventories for the current selector/repair owners, and rewrote pagination stress-page updates to remove/insert stress blocks instead of calling tx.value.replace.
Broad-gate focused verification passed: cd .tmp/slate-v2 && bun --filter slate-react test:vitest -- test/kernel-authority-audit-contract.test.ts test/surface-contract.test.tsx -> 2 files passed, 39 tests passed.
Broad-gate focused typecheck passed: cd .tmp/slate-v2 && bun --filter slate-react typecheck.
Broad-gate site typecheck passed: cd .tmp/slate-v2 && bun typecheck:site.
Broad final gate passed: cd .tmp/slate-v2 && bun check -> lint passed, package/site/root typecheck passed, 1260 Bun tests passed, 37 slate-layout tests passed, and 54 slate-react Vitest files passed with 483 tests.
Final autoreview passed after broad-gate fixes: cd .tmp/slate-v2 && /Users/zbeyens/git/plate-2/.agents/skills/autoreview/scripts/autoreview --mode local -> clean, no accepted/actionable findings.
Issue/reference sync unchanged at closeout: implementation and coverage changes are harvested substrate work only and make no public issue claim.
Plan hygiene and mechanical check passed after execution updates: cd /Users/zbeyens/git/plate-2 && node .agents/rules/autogoal/scripts/check-complete.mjs docs/plans/2026-05-29-slate-v2-portabletext-harvest-plan.md -> complete.

Findings:

Portable Text is a good source of behavior pressure, not an API model.
The best Slate steals are root-aware selection projection, object-boundary insert/delete, self-drop suppression, raw fragment/clipboard behavior, IME transport, remote selection rebasing, and operation/path determinism.
The worst move would be copying Portable Text schema gates or behavior API names into raw Slate. That would make Slate opinionated in the wrong layer.
PT-H12 is useful pressure for transaction/effect boundaries, but only after an explicit slate-plan API decision. Do not sneak a public behavior API in through tests.

Decisions and tradeoffs:

Run PT-SV2-Q1 first because recent bugs clustered around multi-root projection and selection traversal.
Keep PT-H02, PT-H06, PT-H07, PT-H10, and PT-H12 split instead of pretending mixed rows are all raw Slate.
Treat PT-P01 and PT-P02 as Plate backlog, not Slate debt.
Prefer refactoring existing Slate v2 tests before adding new files because current coverage is already broad and scattered.
Require browser proof for drag/drop, clipboard DOM, native selection, and IME claims. Package tests alone are not enough there.

Error attempts:

Error / failed attempt	Count	Different move	Resolution
Scratchpad helper generated the older UTC-dated path and refused to overwrite `docs/plans/2026-05-28-slate-v2-portabletext-harvest-plan.md`.	1	Keep the older plan intact and write the required `2026-05-29` path explicitly.	New dated plan created at the objective path.
Older plan counted 16 behavior rows and stale 211/1979 test-index numbers.	1	Use the fresh 2026-05-29 harvest report and source-routing appendix.	This plan uses 14 families, 200 indexed portable/mixed files, and 1944 anchors.

Verification evidence:

test -f docs/editor-test-harvester/portabletext/inventory.md && test -f docs/editor-test-harvester/portabletext/test-index.md && test -f docs/editor-test-harvester/portabletext/source-routing.md returned harvest-files-ok.
rg --files over .tmp/slate-v2 found current owner anchors for selection, roots, voids, clipboard, drag/drop/root interaction, collab/history, IME, focus/keyboard, operations, and transaction contracts.
node -e "const p=require('./.tmp/slate-v2/package.json'); ..." confirmed bun check, bun check:full, focused package tests, slate-browser tests, and integration-local commands exist.
Required section grep passed for Harvest grounding, Lane contract, Full harvest row accounting, In-lane candidate matrix, Execution queue, Downstream lane application, and Accepted-plan execution handoff.
Placeholder hygiene grep passed with no template markers, progress-state markers, or placeholder-only lines.
Autogoal mechanical completion passed: [autogoal] complete: docs/plans/2026-05-29-slate-v2-portabletext-harvest-plan.md.

Reboot status:

Question	Answer
Where am I?	Closure verification for the `slate-v2` Portable Text harvest plan.
Where am I going?	Run section greps, placeholder scan, autogoal completion check, then close the active goal if all requirements pass.
What is the goal?	A dated execution-grade planning-only Slate v2 harvest plan from Portable Text with every row routed and no implementation edits.
What have I learned?	Portable Text strongly pressures raw roots, projection, object boundaries, IME, clipboard, drag/drop, collab/history, and operation semantics; product schema and plugin policy must stay out of raw Slate.
What have I done?	Read the harvest report and routing appendix, searched current `.tmp/slate-v2` owner coverage, wrote row accounting, split decisions, execution queue, and accepted-plan handoff.

Open risks:

Execution risk remains for browser-heavy rows: drag/drop, clipboard DOM, native selection over roots/voids, and IME. This plan names proof routes but does not execute them.
PT-H12 could become API churn if treated as Portable Text behavior API adoption. The plan constrains it to raw transaction/effect ordering only after an accepted slate-plan API decision.
Existing coverage is broad and scattered. The execution pass should refactor or name current contracts before adding duplicate tests.