List-unit fragment proofs should treat list-item fragments as sibling units and assert real paste landings

What happened

The next geometry seam after richer-inline wrapper stacks was not arbitrary trees.

It was a sharper shape:

• one list-item
• one paragraph child
• one nested bulleted-list child

The first proofs exposed two lies:

• nested bulleted-list targets were unwrapping list-item fragments into paragraph children instead of inserting them as sibling units
• browser assertions were still guessing the post-paste landing path instead of checking where the real clipboard/runtime stack actually placed it

What fixed it

The durable seam was:

• treat list-item fragments as sibling units when the target container is a bulleted-list
• prove the shape through:
  • slate-v2 contracts
  • slate-dom-v2 clipboard-boundary tests
  • real Chromium copy/paste example tests

Why this works

This shape is not “more mixed inline”.

The fragment being pasted is itself the sibling unit. If the code unwraps it, the tree lies immediately.

Once the unit stays intact, the remaining browser truth is simple:

• top-level list-unit explicit inserts still rebase into the split sibling unit
• quote-wrapped list-unit nested-list inserts preserve the outer paragraph selection
• real clipboard paste lands at the deepest inserted descendant, not at the first tempting chip leaf

Reusable rule

When a proof inserts list-item fragments into a bulleted-list target:

• treat the fragment as a sibling unit, not inline content
• promote the shape only after core, DOM, and browser all agree
• assert the real post-paste landing point the browser/runtime stack returns

Related issues

• 2026-04-05-wrapper-block-units-with-inner-paragraphs-can-reuse-the-current-block-geometry-model-at-top-level.md
• 2026-04-05-nested-wrapper-stacks-can-be-promoted-once-browser-proof-matches-the-contract-layer.md