Problem
The four review checklists (metadata, data, docs, tests) exist in 3 to 4 diverging copies:
CLAUDE.md (master sections, ~37 checkbox items total)
commands/review-package.md (first issue body template)
commands/review-issue.md (per-issue conditional blocks)
commands/create-next-issue.md (issue body templates for issues 2 to 4)
docs/review-checklist.csv (27 items, referenced by README as "the complete list" but used by nothing, and not matching the markdown items)
The PR body template similarly exists in 3 places with formatting drift.
Risk to manage
Consolidation can silently change the review standard. A wrong pick during merging (dropped item, wrong function name) would let packages pass review while missing checks the old workflow enforced, and defects would surface only in published packages with permanent DOIs. Depends on #4 (fix known contradictions first, so bugs are not canonicalized).
Tasks
Problem
The four review checklists (metadata, data, docs, tests) exist in 3 to 4 diverging copies:
CLAUDE.md(master sections, ~37 checkbox items total)commands/review-package.md(first issue body template)commands/review-issue.md(per-issue conditional blocks)commands/create-next-issue.md(issue body templates for issues 2 to 4)docs/review-checklist.csv(27 items, referenced by README as "the complete list" but used by nothing, and not matching the markdown items)The PR body template similarly exists in 3 places with formatting drift.
Risk to manage
Consolidation can silently change the review standard. A wrong pick during merging (dropped item, wrong function name) would let packages pass review while missing checks the old workflow enforced, and defects would surface only in published packages with permanent DOIs. Depends on #4 (fix known contradictions first, so bugs are not canonicalized).
Tasks
docs/so every dropped or altered item is on record