[aglinxinyuan]

Summary

• send the Antigravity app's hub client identity on the quota-summary request so AI Pro accounts receive the separate Gemini 4 Argon weekly group
• query the Daily summary host first and move to the next host on any rejection except 401
• limit groups named after one model family (Gemini 4 Argon) to that family in auto-switch, recommendations and weekly warmup

Validation

• 6 focused unit test files (81 tests) and the full unit suite (3354 passed), run locally with vite 7 because tests-unit fails on main (vitest/vite mismatch)
• live AI Pro account on Windows: the weekly section shows Gemini 4 Argon
• TypeScript type-check
• targeted ESLint

Before	After

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota.

[codereviewbot-ai]

Summary & Review Outcome: LGTM

The changes address the missing "Gemini 4 Argon" weekly quota group in Google AI Pro accounts cleanly and safely.

Key Highlights Verified:

1. Hub Client Identity & Version Resolution (request-user-agent.ts, GoogleAPIService.ts):
   • buildHubUserAgent and resolveHubClientVersion correctly implement the required hub identity format (antigravity/hub/<version> (aidev_client; os_type=<os>; arch=<arch>)) with a version floor of 2.26.0.
   • The hub identity is isolated exclusively to GoogleAPIService.fetchQuotaSummary; other Cloud Code calls retain the standard client user-agent.
2. Quota Summary Host Fallback & Retry Logic:
   • QUOTA_SUMMARY_ENDPOINTS prioritize daily-cloudcode-pa.googleapis.com ahead of production and sandbox.
   • HTTP 403 with project retries once without project on the same host before moving on.
   • HTTP 401 terminates immediately (rejected credential), whereas other host-specific non-2xx errors gracefully advance to the next endpoint without dropping model quota.
3. Dedicated Quota Group Scoping (quota-groups.ts, quota-model-families.ts):
   • getDedicatedQuotaGroupFamilyId matches registered product families by exact normalized display name.
   • Dedicated groups (such as Gemini 4 Argon) match only tokens of that specific model family, preventing general gemini match tokens from inappropriately bounding other Gemini models (Flash/Pro) by the Argon limit.
4. Impact on Auto-Switch & Weekly Warmup (AutoSwitchService.ts, WeeklyWarmupService.ts):
   • AutoSwitchService.isAccountDepleted verifies whether any enabled model matches a depleted dedicated quota group family. Because gemini-4-argon-* is not an active proxy model, an exhausted Argon weekly bucket does not trigger account depletion switches for other Gemini models.
   • WeeklyWarmupService.selectWeeklyWarmupCandidates restricts warmup candidates to groups whose dedicated family matches the representative model, preventing warmup requests from being wasted on Argon buckets.
5. Quality & Documentation:
   • Agent note and module documentation reflect current architecture, decisions, and failure behavior.
   • Comprehensive unit tests cover all new behaviors and edge cases.

---

🤖 Reviewed by codereviewbot.ai - Catch bugs before your team does.

[codereviewbot-ai]

Review Summary (LGTM)

The pull request cleanly addresses fetching and handling of the Gemini 4 Argon quota group:

1. Client Identity & Endpoint Order:

   • request-user-agent.ts introduces buildHubUserAgent and resolveHubClientVersion, correctly maintaining the minimum hub version floor (2.26.0) with semver comparison.
   • GoogleAPIService.fetchQuotaSummary sends the required hub client identity header and prioritizes the Daily host endpoint while gracefully falling back across endpoints for all non-401 responses (preserving 403 project-stripping retries).

2. Dedicated Quota Group Scoping:

   • quota-model-families.ts and quota-groups.ts correctly register and isolate gemini-4-argon via getDedicatedQuotaGroupFamilyId. Generic provider tokens ('gemini') no longer inadvertently match dedicated family quota groups.
   • AutoSwitchService only flags accounts as depleted for dedicated groups if matching models are enabled.
   • WeeklyWarmupService ensures warmup requests (e.g., using gemini-flash) do not attempt to claim consumption of dedicated quota groups like Gemini 4 Argon.
   • manual-account-recommendation correctly ignores Argon limits when evaluating accounts for Flash/Pro recommendations.

3. Documentation & Tests:

   • Agent note and module documentation are updated in alignment with repository guidelines.
   • High test coverage provided with targeted unit tests across all updated services.

---

🤖 Reviewed by codereviewbot.ai - Catch bugs before your team does.