fix(billing): re-read the usage gate often once a payer nears its limit

[waleedlatif1]

Summary

• The usage gate held a within-limit admission for 5 minutes, so a payer who crossed the limit kept running executions, tools, and copilot steps for up to that long — unbilled spend on free plans, spend past the cap on capped plans
• An admission is now cached per entry by headroom: 10s when the payer or acting member is near its limit (headroom under the larger of $25 and 20% of the limit — every free-plan payer), the full 5 minutes otherwise, so paid pools with room keep today's cache and read load
• Same rule for the direct-v1 account verdict cache used by copilot continuations
• Refusal caching, unreadable-usage handling, and timeouts are unchanged

Type of Change

• Bug fix

Testing

• usage-gate-cache.test.ts: free payer (floor), large pool in its last 20% (share), member near its cap, ample headroom held for the full TTL — the three near-limit cases fail with the fix reverted
• validate/route.test.ts: direct-v1 continuation near the limit is refused after the near-limit window — fails with the fix reverted
• update-cost/route.test.ts passing; biome on changed files; full test/type-check/audits in CI

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing (new tests pass the test-audit authoring gate)
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
docs	Ready	Preview	Oct 15, 2026 10:47pm UTC

[cubic-dev-ai]

No issues found across 6 files

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

_{Turn on auto-fix | Re-trigger cubic}

[greptile-apps]

Confidence Score: 5/5

[High impact] The PR appears safe to merge; no actionable issues remain from this review.

Summary

The PR shortens cached usage admissions near a payer’s or member’s limit.

• Usage checks refresh near-limit admissions after ten seconds.

Diagram

%%{init: {'theme': 'neutral'}}%%
flowchart TD
  A[Fresh usage answer] --> B{Admission allowed?}
  B -->|Yes| C{Payer or member near a limit?}
  C -->|Yes| D[Cache for 10 seconds]
  C -->|No| E[Cache for five minutes]
  B -->|No| F[Keep existing refusal behavior]

Loading

_{Reviews (2) · Last reviewed commit: "test(billing): drive the gate-cache cloc..." · Reviewed by Greptile}

[waleedlatif1]

@greptile

[waleedlatif1]

@cubic-dev-ai review this PR

[cubic-dev-ai]

@cubic-dev-ai review this PR

@waleedlatif1 I have started the AI code review. It will take a few minutes to complete.

[cubic-dev-ai]

No issues found across 7 files

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

_{Turn on auto-fix | Re-trigger cubic}