Goal

Replan the remaining slate-v2 work around the most lossless path.

The working intent is:

• do not start a broad slate rewrite just because tranche 3 says slate
• preserve live behavior and public contract width wherever possible
• cut only tests or proof lanes that are no longer relevant to a live owned surface
• keep anything uncertain in a named lane instead of silently deleting it

Sources To Re-read

• docs/slate-v2-draft/commands/replan-remaining-work.md
• .omx/specs/deep-interview-slate-v2-true-slate-rc-roadmap.md
• docs/slate-v2/master-roadmap.md
• docs/slate-v2/release-readiness-decision.md
• docs/slate-v2/release-file-review-ledger.md
• docs/slate-v2/true-slate-rc-proof-ledger.md
• docs/slate-v2/references/pr-description.md
• docs/plans/2026-04-13-slate-v2-full-no-regression-story-plan.md
• docs/plans/2026-04-13-slate-v2-exhaustive-api-contract-recovery-plan.md

Questions To Settle

1. What is actually still blocking release-readiness after the Bun/TS6 and slate-browser batches?
2. Which remaining rows are:
   • required for lossless truth
   • safe to cut
   • safe to defer as named post RC
3. How much slate package work is still mandatory before any test cuts become honest?

Working Hypothesis

• the old remaining-work story is too rewrite-first
• the next honest batch is probably a classification pass, not broad source recovery
• any lane that still protects public contract width or live proof ownership should stay open
• “non relevant test anymore” must be proven per test family, not used as a blanket excuse to delete pain

Findings

1. The live roadmap regressed to a broader rewrite-first story

Current live docs say tranche 3 is broad slate package recovery, then support packages, then DOM, then React.

That is too coarse for the current state.

It skips the step that the older re-interview and no-regression plans already identified as the real blocker:

• exhaustive API/public-surface claim-width audit
• deleted-family classification against the live claim

2. Lossless does not mean bulk carry-forward

The fresh-branch reset spec already settled the knife-edge definition:

• recover legacy truth
• adopt non-conflicting real value
• park everything else as named post RC

So the next plan cannot be:

• rewrite slate
• then see what broke

That would be louder, not more lossless.

3. Replan hard rule still applies

The replan command is explicit:

• do not collapse remaining work to browser/input only while public-surface contract-width audit is still open
• if a kept helper is narrower than legacy, reopen that lane before planning the next batch

So any “just cut irrelevant tests” plan is incomplete unless it keeps the claim-width audit open.

4. Deletion closure must follow live proof owners

The strongest workflow learnings all point the same way:

• green harness rows do not automatically mean mirrored runtime parity
• example parity is different from narrow behavior proof
• deleted test/example families must be frozen against the current diff and classified by current proof owner
• explicit skip is honest only when the live claim is narrower and that cut is stated

Decisions

1. The next mainline blocker is classification, not broad source recovery

The next batch should:

1. freeze the current remaining deleted/residue inventory
2. audit kept public-surface rows for claim-width drift
3. classify each remaining row as:
   • keep and recover now
   • explicit skip
   • post RC
4. only then start targeted package/source recovery for the rows that still belong in the live claim

2. slate remains first, but only as the first audited owner

slate still goes first because it owns the widest public surface.

But tranche 3 should mean:

• audit + classify slate
• recover only the kept deltas

not:

• broad same-path source rewrite of packages/slate/src/**

3. Test cuts are allowed only as explicit claim decisions

A test or file can be cut only when all of these are true:

• the current live claim is narrower
• the surviving proof owner is named
• the rationale is explicit
• the docs/ledgers update in the same turn

Anything else is just deleting evidence.

4. Example and browser lanes stay owner-based

• narrow browser proof does not close example parity
• benchmark rows do not masquerade as behavior rows
• root proof/benchmark commands stay outside the mainline until their owning package/example claim is live

Proposed Remaining Order

Tranche 3

Lossless core claim-width audit and residue classification in slate.

Tranche 4

Targeted support-package closure in slate-history and slate-hyperscript, following the audited live claim.

Tranche 5

DOM-owned closure in slate-dom, again only for kept seams.

Tranche 6

React-owned closure in slate-react, tied back to source truth instead of browser proof vibes.

Tranche 7

Contributor-facing example parity, benchmark ownership, additive value, and remaining root proof command surface.

Tranche 8

Final RC ledger closure with named post RC leftovers only.

Required Live Doc Sync

• docs/slate-v2/master-roadmap.md
• docs/slate-v2/fresh-branch-migration-plan.md
• docs/slate-v2/release-readiness-decision.md
• docs/slate-v2/true-slate-rc-proof-ledger.md
• docs/slate-v2/replacement-gates-scoreboard.md
• docs/slate-v2/references/pr-description.md
• docs/slate-v2/release-file-review-ledger.md

Landed Sync

• docs/slate-v2/master-roadmap.md
• docs/slate-v2/fresh-branch-migration-plan.md
• docs/slate-v2/release-readiness-decision.md
• docs/slate-v2/true-slate-rc-proof-ledger.md
• docs/slate-v2/replacement-gates-scoreboard.md
• docs/slate-v2/references/pr-description.md
• docs/slate-v2/release-file-review-ledger.md
• docs/plans/2026-04-13-slate-v2-full-no-regression-story-plan.md
• docs/plans/2026-04-13-slate-v2-exhaustive-api-contract-recovery-plan.md

Verification

• re-read the touched roadmap/verdict/ledger docs
• grep-checked the live docs for stale rewrite-first and pre-replan blocker wording
• confirmed the live stack now says:
  • tranche 3 is claim-width audit + residue classification
  • broad package rewrite is not the next default move
  • targeted recovery happens only after kept-surface classification