Custom menu-bar cost tokens stay stale after token-cost refresh

[vincent-peng]

Summary

Cost Today and 30d Cost in a custom menu-bar layout render from token-cost snapshots. Publishing a new token snapshot invalidates menus, but does not enter the menu-bar icon/title refresh path. The displayed cost can therefore remain – or stale until an unrelated usage or settings change occurs.

Source path

• StatusItemController+MenuBarLayout reads the current provider token snapshot to render Cost Today and 30d Cost.
• UsageStore.menuObservationToken observes token snapshots and causes menu invalidation.
• UsageStore.iconObservationToken and the corresponding icon-observation signature do not include token snapshots or their cost values, so they do not request a title refresh for a cost-only update.

Reproduction

1. Save a custom menu-bar layout containing Cost Today or 30d Cost.
2. Keep the provider usage snapshot unchanged.
3. Publish or refresh a token-cost snapshot.

Expected behavior

The rendered menu-bar title should update when the cost value used by the active layout changes.

Suggested acceptance criteria

• Add focused coverage that publishes a token-cost snapshot while the usage snapshot is unchanged and asserts that the rendered title/invalidation changes.
• Include only render-relevant token-cost state in menu-bar refresh observation, avoiding broad invalidation for unrelated token data.

[clawsweeper]

Codex review: keeping this open for maintainer follow-up; there is still a little grit to resolve. Reviewed July 18, 2026, 10:03 AM ET / 14:03 UTC.

Summary
Keep this issue open. The report gives a narrow, credible stale-title path with a focused acceptance boundary, but current-main source and history verification is still needed before treating it as fixed or dispatching an automatic repair.

Reproducibility: no. current-main reproduction was completed in this review. The report provides concrete steps and a precise source-level hypothesis, so a focused regression test should be able to confirm it quickly.

Ways to help us reproduce this

• Add a screenshot or short recording showing the behavior.
• Add expected vs actual behavior.
• Include redacted logs or terminal output.

Next step
The report names a bounded observation-to-render gap and a focused regression scenario; verify the exact current fields and test seam before implementing.

Review details

Best possible solution:

Update the menu-bar refresh observation to include only the token-cost values consumed by the active cost layout, with a regression test proving a cost-only publication changes the rendered title or its invalidation state.

Do we have a high-confidence way to reproduce the issue?

No current-main reproduction was completed in this review. The report provides concrete steps and a precise source-level hypothesis, so a focused regression test should be able to confirm it quickly.

Is this the best way to solve the issue?

Likely yes. Restricting title-refresh observation to the token-cost state actually rendered by the active layout is narrower and safer than broadly invalidating on every token snapshot change.

AGENTS.md: found and applied where relevant.

Remaining risk / open question:

• The exact current observation signature, token-cost fields, and existing XCTest seam need source verification before changing invalidation behavior, so unrelated token updates are not made title-refresh triggers.

Codex review notes: model internal, reasoning high; reviewed against 2ccb4525687c.

Label changes

Label changes:

• add P2: A custom menu-bar layout can display incorrect or stale cost information, but the report describes a bounded refresh defect without evidence of data loss or an unavailable core runtime.
• add impact:ux-friction: Users can continue using the app, but cost values in a configured menu-bar layout may remain misleading until an unrelated refresh occurs.
• add issue-rating: 🦪 silver shellfish: Current issue advisory state selects this label.

Label justifications:

• P2: A custom menu-bar layout can display incorrect or stale cost information, but the report describes a bounded refresh defect without evidence of data loss or an unavailable core runtime.
• impact:ux-friction: Users can continue using the app, but cost values in a configured menu-bar layout may remain misleading until an unrelated refresh occurs.

Evidence reviewed

Acceptance criteria:

• Add and run a focused XCTest covering unchanged usage with an updated token-cost snapshot and changed menu-bar title or icon-observation state.
• make test
• make check

What I checked:

• Concrete report and reproduction: The report identifies a three-step path: retain an unchanged provider usage snapshot, publish a token-cost snapshot, and observe that a custom cost-based title can remain stale. (Sources/CodexBar/StatusItemController+MenuBarLayout.swift)
• Narrow observation boundary: The requested change is limited to render-relevant token-cost state in the icon/title observation signature, while menu invalidation already observes token snapshots according to the supplied report. (Sources/CodexBar/UsageStore.swift)
• Repository testing policy: The applicable repository guidance calls for focused XCTest coverage for parser/provider-style logic and requires make test plus make check after a code change; it also favors stable state/model seams over live AppKit menu construction. (AGENTS.md)

Likely related people:

• vincent-peng: The report supplies the relevant rendering and observation surfaces plus a bounded regression scenario; no current-main history attribution could be established in this review. (role: reporter and source-path investigator; confidence: low; files: Sources/CodexBar/UsageStore.swift, Sources/CodexBar/StatusItemController+MenuBarLayout.swift)

How this review workflow works

• ClawSweeper keeps one durable marker-backed review comment per issue or PR.
• Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
• A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
• PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
• Maintainers can also comment @clawsweeper review to request a fresh review only.
• Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge.
• Maintainer-only repair and merge flows require explicit commands such as @clawsweeper autofix, @clawsweeper automerge, @clawsweeper fix ci, or @clawsweeper address review.
• Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.

[clawsweeper]

🦞🔧
ClawSweeper is automatically building this issue.

The issue review finished and no active implementation pull request was found.

Automatic implementation progress:

• State: Queued
• Detail: Review complete; no active implementation PR was found. The Codex implementation worker is queued.
• Run: https://github.com/openclaw/clawsweeper/actions/runs/29649522903
• Updated: 2026-07-18T15:13:07.111Z
• Opt out: add clawsweeper:manual-only or clawsweeper:human-review.

[steipete]

Fixed by #2305, merged as 841de0de552c0ea3d2efa90e15a287bd5f42c167.

The custom menu-bar icon observation now follows token-cost snapshot publications and includes the rendered costToday/cost30d value only when that token is active. Focused coverage passed 17/17 tests, make check passed, final autoreview found no issues, and exact-head CI run 29655359387 is fully green.