Coverage Priority Map Post Yjs Merge

Goal

Run fresh repo coverage, score every packages/*/src/** file for non-React unit-test value, and sync the next recommendations against:

• March 6, 2026 cleanup strategy
• March 9, 2026 excellence plan
• March 17, 2026 execution passes
• the now-merged Yjs suite work

Coverage Run

• Command:
  • bun test --coverage --coverage-reporter=lcov --coverage-dir=.coverage-repo-2026-03-22b --reporter=dots
• Result:
  • 2557 pass
  • 0 fail
  • 444 files
  • 4.25s
• Artifact:
  • lcov.info

Sync From Completed Work

• March 14, 2026:
  • markdown already got the worthwhile helper and fallback pass.
• March 15, 2026:
  • table already got the merge, sizing, and selection helper pass.
• March 17, 2026:
  • selection, docx-io, and the focused non-React core lane already got follow-up work.
• March 22, 2026:
  • yjs just got the fast suite, slow collaboration lane, and follow-up upstream-inspired cases.

Strong take: the raw matrix still slightly over-scores core and table because they are big and still have uncovered leftovers. I would not reopen them next. The actual recommendation order below is the one to follow.

Scoring Rules

• Scope is every runtime file under packages/*/src/**.
• /react files score 0 by design.
• Barrels, declaration files, and type-only files score 0.
• Declarative rule tables under autoformat/rules/** score 0.
• Recent package passes get penalized so we do not recommend the same work twice.
• Scores favor deterministic transforms, parsers, serializers, queries, provider logic, plugin contracts, and pure helpers with meaningful uncovered behavior.
• UI-skewed packages get pushed down even when they are uncovered.

My Order

1. list-classic

Best next package.

• Why:
  • best remaining mix of deterministic query and transform seams
  • high-value editor behavior, not coverage cosplay
  • sits exactly where the older plan already wanted the next serious spend
• Start here:
  • getTodoListItemEntry.ts
  • getHighestEmptyList.ts
  • withDeleteFragmentList.ts
  • withInsertBreakList.ts
  • moveListSiblingsAfterCursor.ts

2. suggestion

• Why:
  • deterministic diff and suggestion-transform behavior
  • strong value without any /react detour
  • better next seam quality than reopening core leftovers
• Start here:
  • diffToSuggestions.ts
  • withSuggestion.ts
  • removeNodesSuggestion.ts
  • getSuggestionNodeEntries.ts
  • findSuggestionNode.ts

3. docx

• Why:
  • very unit-testable cleaner and import seams
  • big user-facing value without dragging app wiring into package tests
  • cleaner next move than docx-io, which already got its pure-helper pass
• Start here:
  • cleanDocxImageElements.ts
  • getRtfImageHex.ts
  • getRtfImagesMap.ts
  • getRtfImagesByType.ts
  • getRtfImageMimeType.ts

4. ai

• Why:
  • the value here is in pure transforms and prompt helpers, not model fakery
  • coverage is still brutally low in the exact deterministic seams that matter
  • worth doing after suggestion and docx, not before them
• Start here:
  • withAIBatch.ts
  • getEditorPrompt.ts
  • insertAINodes.ts
  • replacePlaceholders.ts
  • undoAI.ts

5. layout

• Why:
  • still mostly deterministic column operations
  • real regression surface
  • no /react requirement
• Start here:
  • resizeColumn.ts
  • moveMiddleColumn.ts
  • insertColumnGroup.ts
  • withColumn.ts
  • insertColumn.ts

6. list

• Why:
  • still has clean query and toggle seams left
  • smaller and cheaper than the packages above
  • good follow-up once list-classic is moving
• Start here:
  • getSiblingListStyleType.ts
  • toggleListByPath.ts
  • someList.ts
  • someTodoList.ts
  • setListNode.ts

7. code-block

• Why:
  • remaining non-React logic is still real
  • cleaner than reopening core or table
• Start here:
  • withCodeBlock.ts
  • formatter.ts
  • withInsertDataCodeBlock.ts
  • setCodeBlockToDecorations.ts

8. autoformat

Surgical only.

• Worth testing:
  • AutoformatPlugin.ts
  • autoformatBlock.ts
  • autoformatMark.ts
  • autoformatText.ts
• Not worth the time:
  • rule-table files under packages/autoformat/src/lib/rules/**

9. csv

Tiny but sharp.

• Start here:
  • deserializeCsv.ts
  • CsvPlugin.ts

10. comment

Good file-level value, but not a bigger campaign than the packages above.

• Start here:
  • BaseCommentPlugin.ts
  • withComments.ts
  • getCommentKeys.ts
  • getCommentCount.ts

Not Next

• yjs
  • just landed a serious fast plus slow pass
  • current matrix score is 0
• media
  • current non-/react matrix score is 0
  • the old March 17 recommendation is stale here
• markdown
  • down to one score-1 leftover on defaultRules.ts
  • not worth the next cycle
• selection
  • only one real leftover file still registers
  • March 17 already consumed the good internal seams
• docx-io
  • pure-helper pass already happened on March 17
  • remaining debt is lower-value export/app crossover
• table
  • still has leftovers, especially withSetFragmentDataTable.ts
  • still not worth reopening right now
• core
  • still has low-coverage leftovers
  • most of them are not as clean or as urgent as the packages above

Ranking Snapshot

This is my actual recommendation order, not just the raw heuristic sort.

Rank	Package	Score	Take
1	list-classic	49	Best next deterministic editor-behavior lane
2	suggestion	46	Rich transform/diff value without /react
3	docx	45	Very unit-testable cleaner/import seams
4	ai	47	Good pure transform/helper lane, avoid network theater
5	layout	44	Strong column transform seams
6	list	41	Cheap query/toggle follow-up
7	code-block	36	Real non-React behavior still left
8	autoformat	25	Surgical pass only
9	csv	19	Two-file slice, clean ROI
10	comment	26	Good file-level value, smaller overall win

Artifacts

• lcov.info
• 2026-03-22-coverage-priority-packages-post-yjs.tsv
• 2026-03-22-coverage-priority-files-post-yjs.tsv

The TSVs are the exhaustive matrix for every package and file under packages/*/src/**. The markdown is the actual call on what to do next.