fix(antigravity): humanize raw model id labels and add localization fallback

[bcharleson]

Summary

Antigravity quota responses can expose raw model identifiers as display labels. This change humanizes generated labels while preserving custom server-provided labels.

• Convert hyphenated identifiers into readable title case.
• Preserve dotted versions and combine consecutive single-digit version components.
• Preserve known acronyms such as GPT, AI, API, and OSS.
• Leave custom labels unchanged.
• Keep empty identifiers empty.
• Apply generated labels only when the server label is empty or matches the raw identifier.
• Add changelog credit for @bcharleson.

The original unrelated localization fallback was removed. Compatibility targets the current Antigravity response shape only.

Public model gate

Test fixtures use publicly documented Google model identifiers such as gemini-3-pro-preview and gemini-2.5-flash. No unannounced model identifiers are included.

Proof

Exact rebased head: c6e9d4faf97b66ec9364816e41a04f84b52bfefa

• 62 focused Antigravity probe and label tests pass.
• make check passes.
• Autoreview: no actionable findings, confidence 0.86.
• Live authenticated CLI probe: app source returned primary and secondary usage plus four extra model windows; every extra window had a nonempty humanized title; no probe error.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 1f9ea14422

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Apply suffix fallback before the English guard

When the app is already using en.lproj or the English bundle cannot be resolved, the guard above exits via return trimmed.isEmpty ? key : value before this new fallback is reached. That means the intended safety net still does not cover the default-English/missing-English-key case: missing usage_percent_suffix_left or usage_percent_suffix_used will continue to render the raw key instead of left/used. Consider applying codexBarLocalizationFallback(for:) in the guard's else path as well.

Useful? React with 👍 / 👎.

[clawsweeper]

Codex review: needs maintainer review before merge. Reviewed June 18, 2026, 12:07 PM ET / 16:07 UTC.

Summary
The PR humanizes Antigravity raw model-ID labels for distinct quota windows, preserves custom server-provided labels, adds focused label tests, updates one status-probe test fixture, and adds a changelog line.

Reproducibility: yes. source inspection gives a high-confidence path by constructing an Antigravity snapshot with a distinct quota whose label == modelId, then observing current main uses quota.label verbatim for extra-window titles. I did not run live provider probes because AGENTS.md warns against unrequested account or Keychain-adjacent validation.

Review metrics: 2 noteworthy metrics.

• Diff surface: 4 files changed; 74 additions, 5 deletions. The patch is small and concentrated on one Antigravity display-label path plus tests and changelog text.
• Focused tests: 1 test file added; 1 Antigravity status test updated. The new coverage directly exercises raw-ID humanization and custom-label preservation.

Merge readiness
Overall: 🐚 platinum hermit
Proof: 🦞 diamond lobster
Patch quality: 🐚 platinum hermit
Result: ready for maintainer review.

Overall follows the weaker of proof and patch quality, so missing proof can cap an otherwise strong patch.

Rank-up moves:

• [P2] Wait for the remaining exact-head CI shards to complete before merge.

Risk before merge

• [P1] GitHub reported the PR as mergeable but UNSTABLE because exact-head macOS CI shards were still queued or running; merge should wait for required checks to finish.

Maintainer options:

1. Decide the mitigation before merge
   Land this focused fallback after exact-head CI passes so distinct Antigravity quota windows show readable generated labels only when the backend did not provide a custom label.
2. Pause or close
   Do not merge this PR until maintainers decide whether the risk is worth taking.

Next step before merge

• [P2] No repair lane is needed because the branch is coherent, proof-backed, and has no actionable review findings; remaining work is normal maintainer merge gating and CI completion.

Security
Cleared: Cleared: the diff changes Swift label formatting, focused tests, and changelog text only, with no dependency, workflow, credential, signing, download, or package-resolution changes.

Review details

Best possible solution:

Land this focused fallback after exact-head CI passes so distinct Antigravity quota windows show readable generated labels only when the backend did not provide a custom label.

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

Yes: source inspection gives a high-confidence path by constructing an Antigravity snapshot with a distinct quota whose label == modelId, then observing current main uses quota.label verbatim for extra-window titles. I did not run live provider probes because AGENTS.md warns against unrequested account or Keychain-adjacent validation.

Is this the best way to solve the issue?

Yes: the PR is the narrow maintainable fix because it only generates a readable label for empty/raw labels and preserves backend-provided custom labels. The merged backend display-name PR remains intact and only partially overlaps this fallback.

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning high; reviewed against 73b35b0f8d2e.

Label changes

Label justifications:

• P2: This fixes a visible Antigravity quota-label bug with limited blast radius to one provider and a narrow display-formatting path.
• rating: 🐚 platinum hermit: Overall readiness is 🐚 platinum hermit; proof is 🦞 diamond lobster and patch quality is 🐚 platinum hermit.
• status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR. Sufficient (live_output): Sufficient: the PR body and owner exact-head comments provide after-fix live CLI probe output showing extra Antigravity model windows with nonempty humanized titles, plus focused tests and make check.
• proof: sufficient: Contributor real behavior proof is sufficient. Sufficient: the PR body and owner exact-head comments provide after-fix live CLI probe output showing extra Antigravity model windows with nonempty humanized titles, plus focused tests and make check.

Evidence reviewed

What I checked:

• Repository policy read: AGENTS.md was read fully before review; its guidance affected the review by keeping validation to source/GitHub evidence instead of live credential or Keychain-adjacent probes. (AGENTS.md:1, 73b35b0f8d2e)
• Current main still uses raw quota labels: On current main, distinct Antigravity extra rate windows set title: m.quota.label, so a quota whose label equals a raw model ID still displays that raw ID. (Sources/CodexBarCore/Providers/Antigravity/AntigravityStatusProbe.swift:561, 73b35b0f8d2e)
• Latest release has the same raw-label behavior: The latest release tag v0.36.1 also maps distinct Antigravity extra-window titles directly from m.quota.label, so the PR is not already shipped in a release. (Sources/CodexBarCore/Providers/Antigravity/AntigravityStatusProbe.swift:575, 06fea2c897cc)
• PR implements a narrow display fallback: The diff adds quotaDisplayLabel and humanizedModelID, using generated labels only when the server label is empty or equals the raw model ID, and changes distinct-window titles to use that helper. (Sources/CodexBarCore/Providers/Antigravity/AntigravityStatusProbe.swift:356, 1c4b46b9519b)
• Focused regression coverage added: The PR adds AntigravityModelLabelTests for hyphenated IDs, dotted versions, consecutive single-digit versions, acronyms, empty IDs, and custom-label preservation. (Tests/CodexBarTests/AntigravityModelLabelTests.swift:1, 1c4b46b9519b)
• Real behavior proof and checks: The PR body and owner comments report exact-head focused tests, make check, release CLI builds, and live authenticated CLI probes returning extra model windows with nonempty humanized titles; GitHub showed several checks successful with macOS shards still queued or running. (1c4b46b9519b)

Likely related people:

• steipete: Current blame on the extra-window title path points to the v0.36.1 release import under Peter Steinberger, shortlog shows the heaviest recent ownership across the relevant files, and the exact-head rebase/proof comments were posted by steipete. (role: recent area contributor and proof owner; confidence: high; commits: 06fea2c897cc, 1c4b46b9519b; files: Sources/CodexBarCore/Providers/Antigravity/AntigravityStatusProbe.swift, Tests/CodexBarTests/AntigravityStatusProbeTests.swift, CHANGELOG.md)
• guhyun9454: The Antigravity per-model quota windows and noisy model-quota filtering that created this distinct-window display surface trace to commits authored by Kwon GuHyeon. (role: feature-history contributor; confidence: high; commits: 352f0b19cfb5, 5dec44e72c02; files: Sources/CodexBarCore/Providers/Antigravity/AntigravityStatusProbe.swift, Tests/CodexBarTests/AntigravityStatusProbeTests.swift)
• Yuxin-Qiao: Yuxin-Qiao authored the merged Antigravity backend display-name update in the same status-probe and test area, which partially overlaps this raw-ID label fallback. (role: recent adjacent contributor; confidence: medium; commits: 70f37afcb522; files: Sources/CodexBarCore/Providers/Antigravity/AntigravityStatusProbe.swift, Tests/CodexBarTests/AntigravityStatusProbeTests.swift)

What the crustacean ranks mean

• 🦀 challenger crab: rare, exceptional readiness with strong proof, clean implementation, and convincing validation.
• 🦞 diamond lobster: very strong readiness with only minor maintainer review expected.
• 🐚 platinum hermit: good normal PR, likely mergeable with ordinary maintainer review.
• 🦐 gold shrimp: useful signal, but proof or patch confidence is still limited.
• 🦪 silver shellfish: thin signal; proof, validation, or implementation needs work.
• 🧂 unranked krab: not merge-ready because proof is missing/unusable or there are serious correctness or safety concerns.
• 🌊 off-meta tidepool: rating does not apply to this item.

Shiny media proof means a screenshot, video, or linked artifact directly shows the changed behavior. Runtime, network, CSP, and security claims still need visible diagnostics.

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.

[steipete]

Post-merge rebase proof for exact head 5ac56927:

• conflict was changelog-only; retained both landed Windsurf and Antigravity entries
• Sources/Tests diff SHA-256 is byte-identical to reviewed head c6e9d4fa
• 62 focused Antigravity model-label/status-probe tests passed
• exact release CodexBarCLI built successfully
• authenticated exact-head live probe: source app, primary and secondary usage present, four additional rate windows, identity metadata present, no provider error
• public model identifier gate remains PASS: model-bearing diff is byte-identical to the previously audited candidate; no unknown identifier was introduced by the rebase

Exact-head CI is running.

[steipete]

Final post-pace rebase proof for exact head 5acca2ae:

• rebase applied without conflict
• Antigravity source/test diff SHA-256 remains byte-identical to the fully reviewed candidate
• exact release CLI rebuilt; authenticated live probe again returned app source, primary/secondary usage, four additional windows, identity metadata, and no error
• public model identifier gate remains PASS; model-bearing diff did not change

Exact-head CI restarted.

[steipete]

Workflow-only final rebase: exact head is now 1c4b46b9. Antigravity source/test diff SHA-256 remains identical to the reviewed/live-proven candidate; the only new base change is landed CI sharding. Faster exact-head CI is queued.