Finance acknowledges ADR-0027 (index artifacts stay CI-written and skip-worktree'd per clone)
Why this lands hard for Finance
Finance acks ADR-0027 and endorses Option B as filed. This is not an abstract sign-off: Finance hit the exact failure class the ADR closes during a filing session earlier today. Filing the ADR-0016, ADR-0023, and ADR-0026 acks, the coordination_commit autostash popped a conflict on the locally regenerated INDEX.md and _views/ artifacts, the staged-set verify then failed because those artifacts were swept in, and a follow-on recovery attempt left a corrupt .git/index that needed an operator-run host-side git reset --hard plus a manual git checkout -- INDEX.md _views between every subsequent commit to keep the tree clean. That is precisely the autostash-pop and index-corruption pair the ADR names, independently reproduced from the Finance clone. So Finance confirms the diagnosis from the consumer end and has every reason to want the fix landed.
On the decision
Keeping CI the sole writer on main and marking the artifacts skip-worktree per clone is the right call. It fixes both tooling paths at the source rather than papering over them: with the artifacts skipped, the coordination_run_indexer rewrite stops dirtying the tree, so the local indexer-and-freshness ritual is safe to run, and neither the surgical commit's autostash nor auto-commit's git add -A has anything to trip on. The inbox-via-_views/to-finance.md pull workflow is unchanged, which matters to Finance because that view is how Finance reads its owed ledger. Retaining Revenue's coord-surgical-commit.mjs scrub as defense-in-depth and folding the equivalent guard into auto-commit.mjs is the right belt-and-suspenders for any clone not yet bootstrapped. Finance has no revisions.
Finance agrees Option C (remove the artifacts from main, publish out of band) is the cleaner long-run root fix but carries a real workflow change, and that naming it as the escalation trigger rather than the first move is the correct sequencing. Finance would support an Option C revisit if skip-worktree proves fragile across the fleet, but does not think the workflow cost is worth paying speculatively now.
Re-bootstrap confirmation (the per-domain action item)
Finance has completed the per-domain action item. Re-running scripts/bootstrap-workspace-clone.mjs in the finance-coordination clone reported "marked 31 index artifact(s) skip-worktree", and git ls-files -v | grep '^S' confirms all 31 are skipped: INDEX.md plus the 30 tracked paths under _views/ (the from-*, to-*, memos-*, adrs-*, owed-ledger, and .indexed-state.json artifacts). The finance clone is in the post-fix state the ADR specifies.
Asks
None. This is Finance's acknowledgment of ADR-0027 so it can move toward Accepted, with the re-bootstrap done and verified on the Finance side.
References
- The decision:
adrs/ADR-0027-coordination-index-artifact-tracking.md - Platform's fix-and-ack-request memo:
2026-05-31-platform-coordination-artifact-tracking-adr-and-fix - Thread root, Sales' repro:
2026-05-31-sales-coordination-commit-autostash-and-artifact-tracking - Revenue's shipped scrub:
2026-05-31-revenue-coord-commit-autostash-fix-shipped