[dbalders]

Summary

Pins the clock in one server test, reports saved usage of a removed directory only for windows it reaches. This is the exact upstream fix (pingdotgg/t3code#28603, 1280ccd0d2). Test-only; one file, 3 lines added and 1 changed.

Why

The test writes a transcript dated 2026-08-01, deletes its folder, and checks that the saved usage still appears for a date range that reaches it. It read with the real clock. Saved usage for deleted files is kept for 92 days, so since 2026-11-01 10:00 UTC the fixture has been past retention. The service correctly drops it, and the test fails Test Server 1 on any PR based on current main (for example #473). Product behavior is unchanged; the test only ever passed because "now" happened to be close to its fixture date.

With "now" pinned to 2026-08-15, the test keeps checking both behaviors:

• saved usage of a deleted folder is still reported for a date range that reaches it;
• a later date range shows no usage and the source is missing.

Validation

• vp check (format and lint), vp run typecheck via vp run check:changed: all passed, including knip and graphql
• Relevant tests passed: apps/server/src/usage/UsageService.test.ts 40/40. Before this change, the same test failed on dd0d5acfd1 with expected [] to deeply equal [ { day: '2026-08-01', … } ] at line 2470.
• git diff --check
• Native mobile lint: not applicable
• No app run: this changes only a test.

Risk and Recovery

None for users; only a test changes. Revert to restore the previous test.

Release and Deployment Impact

None.

Upstream Alignment

Byte-identical to upstream 1280ccd0d2 (cherry-picked with -x), so the next upstream sync applies cleanly. Upstream's later 8b4de61b8b adds a redundant second setTime before setup; the sync will bring that in.

#474 currently carries its own version of this fix (it.effect pinned to 2026-08-02). Once this lands, #474 can drop that edit and stay UI-only.

Checklist

• The PR is ready for review and is not a draft
• The change is focused and avoids unrelated churn
• New behavior has regression coverage where practical
• No secrets, credentials, private URLs, or sensitive logs are included
• Actionable review findings have been resolved or explicitly answered

Implementation: Claude Opus 5.5 via Claude Code

[coderabbitai]

Warning

Review limit reached

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

Next included review available in 40 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available.

Learn how review limits work.

Review configuration:

Run configuration

• Source mode: Auto
• Repository: dbalders/TritonAI-Harness
• Branch: fix/usage-saved-window-clock-20261102
• File: .coderabbit.yaml
• Commit: a57b752
• Layers: Repository YAML
• Review profile: CHILL
• Plan: Advanced
• Run ID: d3e4899e-1f89-4cb7-ba03-1e3a46f144af

Commits

Reviewing files that changed from the base of the PR and between 4429dcc and a57b752.

Files selected for processing: 1

• apps/server/src/usage/UsageService.test.ts

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

Share

• X
• Mastodon
• Reddit
• LinkedIn

[dbalders]

Cross-model review blocked: authentication

Implementation (PR footer): Claude Opus 5.5 via Claude Code
Reviewer: Codex gpt-6.1-sol (high)
Commit: a57b752e1cd877acad2d595426cc50dcf1a5b1ff

The review did not finish, so no verdict was produced. Reason: ReviewAuthenticationError: codex authentication was rejected during review on the review service. No review verdict was produced.

Sign in to the reviewer on the review service host, then comment /cross-review to retry. For Claude, run claude auth login --claudeai; for Codex, run codex login.

---

Automated cross-model review, updated 2026-11-02T18:25:02+00:00.

[dbalders]

TritonAI Review

Model: api-glm-5.3 · Reasoning: high.

No actionable findings in the inspected changes.

Reviewed a57b752e1cd877acad2d595426cc50dcf1a5b1ff against 4429dcc73fb933cb1c76ecf70997e144f67ce396.

Reviewed 1 file(s) together as one PR.

Changes

• PR: This PR is a minimal, test-only change that pins the clock in one UsageService test (reports saved usage of a removed directory only for windows it reaches). A TestClock.setTime to 2026-08-15 is added after setup, and the test's pipe gains Effect.provide(TestClock.layer()) so the service's retention/pruning logic sees the pinned

CI snapshot

Check runs:
- Sync PR size label definitions: status=completed conclusion=skipped app=
- Label PR size: status=completed conclusion=success app=
- Windows Regression Tests: status=in_progress conclusion=none app=
- Macroscope - Effect Service Conventions: status=completed conclusion=skipped app=
- Macroscope - Correctness Check: status=completed conclusion=skipped app=
- CodeRabbit: status=completed conclusion=neutral app=
- Prepare PR size config: status=completed conclusion=success app=
- Test: status=queued conclusion=none app=
- Test Server 4: status=queued conclusion=none app=
- Test Server 3: status=queued conclusion=none app=
- Test Server 1: status=in_progress conclusion=none app=
- Test Server 2: status=in_progress conclusion=none app=
- Test Server 5: status=queued conclusion=none app=
- Preflight: status=queued conclusion=none app=
- Mobile Native Changes: status=queued conclusion=none app=
- Release Smoke: status=in_progress conclusion=none app=
- Rust: status=completed conclusion=success app=
- Native fingerprint diff: status=completed conclusion=success app=
- Windows Test Scope: status=completed conclusion=success app=

CI is reported separately. This code review does not claim to have run tests.

Comment /triton-review to request a fresh review. Findings are advisory; GitHub branch protection remains authoritative.

[dbalders]

/cross-review

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	5.0 KiB	5.0 KiB	−7 B (−0.1%)	6.8 KiB	✅
Codex	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Codex	Live turn WebSocket wire	1.2 KiB	1.2 KiB	−7 B (−0.6%)	2.0 KiB	✅
Codex	Live turn WebSocket decoded	20.9 KiB	20.9 KiB	0 B (0.0%)	29.3 KiB	✅
Codex	Live turn messages	2	2	0 (0.0%)	8	✅
Claude	Total thread wire	5.0 KiB	5.0 KiB	+7 B (+0.1%)	6.8 KiB	✅
Claude	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Claude	Live turn WebSocket wire	1.2 KiB	1.2 KiB	+7 B (+0.6%)	2.0 KiB	✅
Claude	Live turn WebSocket decoded	21.2 KiB	21.2 KiB	0 B (0.0%)	29.3 KiB	✅
Claude	Live turn messages	2	2	0 (0.0%)	8	✅

Baseline: 4429dcc · PR result: a57b752 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 108.5 KiB
• Claude decoded thread snapshot: 108.8 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[dbalders]

Note

🤖 Claude Opus 5.5 responding on behalf of David

Closing as superseded: #474 (squash 4429dcc) already put a clock fix for this test on main (it.effect, clock set to 2026-08-02T12:00Z). David chose to keep that version. The next upstream sync will hit a 3-line conflict here against upstream 1280ccd0d2; resolve it by taking upstream's version. The branch is kept.