[teamleaderleo]

Codex excursion, unverified.

Summary

Persisted chat attachments now collapse duplicate attachment IDs by keeping the latest chip. Previously, duplicate IDs wrote multiple files but retained duplicate list entries pointing at the final file, so reopening a draft duplicated content and orphaned the earlier file.

Validation

• git diff --check
• Added ChatDraftAttachmentsTests.duplicateAttachmentIdsKeepOnlyTheLatestChip.
• Swift tests and macOS app builds were not run locally; use the fleet compile-check.

No capture: persistence-only change.

Agent session: cosmic-gray

---

^{Need help on this PR? Tag @codesmith-bot with what you need.}

---

Summary by cubic

Persisted chat attachments now collapse duplicate attachment IDs by keeping the latest chip. Previously, duplicate IDs wrote multiple files but left duplicate list entries pointing at the final file, so reopening a draft duplicated content and orphaned the earlier file.

The deduplication treats the latest entry as authoritative, matching the composer's current list semantics. Added ChatDraftAttachmentsTests.duplicateAttachmentIdsKeepOnlyTheLatestChip to cover the behavior.

^{Written for commit 94ca9ad. Summary will update on new commits.}

[github-actions]

cmux status _{(one comment, edited in place; each line is updated by the job that owns it)}

Review 94ca9ad: HQ feed card · gallery before/after runs · checks

App of this head (app.zip, kept 1 day)

gh workflow run reload-build.yml -R manaflow-ai/cmux --ref feat-cmux-next -f tag=pr23394 -f ref=codex-x/edge-cases/huge-paste -f platform=macos -f backend_mode=local

Not verified: macOS jobs did not run: compile and app tests were skipped on 94ca9adc7f; ci-status is green without them.

CI passes on 94ca9adc7f (run 39602529249 attempt 1).
macOS jobs did not run: compile and app tests were skipped: Fast static checks skipped (the macos job needs both to succeed). Re-run CI once they pass, before merging.

Written by scripts/ci/classify_failures.py (ci-failure-attribution.yml); signatures are its SIGNATURES table. A machine verdict is the runner's fault, not this PR's; yours means the failing file is one this PR changes, also red on main that main's latest full suite fails the same way, seen on other PRs that it failed on another pull request's run lately.

[teamleaderleo]

Bead cx-9mq8r created for this PR by the cmux coordinator.

[teamleaderleo]

Review (excursion-verifier subagent, correctness first): SHIP at 94ca9ad

• moveOut keeps the last chip per id in original order; nonisolated, pure locals; both callers (draft write, relaunch save) only read files/bytes/attachments, no count or index reliance.
• Test goes keep -> write -> moveOut -> readBack; without the fix kept ids would be ["same","same"], both with chip 2 bytes, so it fails. Helpers and Chip init match; keep is @discardableResult.
• Note: byte-carrying chips get crypto.randomUUID and wake/kept-read/attach already dedupe by id, so this is defensive. The one live duplicate source is link chips (link:<path>) appended without dedupe in composer/attachmentIntake.tsx:80; worth a source fix. The comment's "first file orphaned" overstates it: both callers sweep unnamed files.

[teamleaderleo]

Owned by claude-sr2-1 on cosmic-gray · session

[github-actions]

Shipped in:

• cmux NEXT nightly 1.0.0-nightly.3960894798201 (0b1a93e9c0)
• cmux NEXT nightly 1.0.0-nightly.3961101575901 (101cc33db2)
• cmux NEXT nightly 1.0.0-nightly.3961563706401 (5c92af7a3b)