[AI-3386] Write the Kiro Crew test fixtures culture-invariantly

[realtonyyoung]

Closes #1250 — AI-3386

What & why

KiroCrewParentResolverTests interpolated a double straight into the JSON it seeds, so under a comma-decimal locale (nb_NO, for one) it wrote 1790266768,153635, which is invalid JSON. The reader then saw no sub-agent record and six tests failed. That fixture now formats the number with CultureInfo.InvariantCulture. KiroCrewSessionEndWatchTests had the same exposure in a custom date format, whose : is the culture's time separator, and now formats invariantly too. Production parsing was already culture-invariant.

Verification

Under LC_ALL=nb_NO.UTF-8, six of the fourteen resolver tests failed before the change. After it, every Kiro Crew test in the core unit suite passes under both nb_NO and en_US.

🤖 Generated with Claude Code

[linear-code]

AI-3386

[qodo-code-review]

PR Summary by Qodo

Make Kiro Crew test fixtures culture-invariant

🐞 Bug fix 🧪 Tests 🕐 Less than 10 minutes

AI Description

• Format sub-agent start times as valid JSON numbers regardless of the machine's locale.
• Format session timestamps consistently so watcher tests work under comma-decimal locales.

Diagram

graph TD
  ParentTests["Parent resolver tests"] --> Json["Fixture JSON"] --> Reader["Kiro Crew records"] --> Resolver["Parent resolver"]
  WatchTests["Session watch tests"] --> Json
  Reader --> Watch["Session end watch"]

Loading

High-Level Assessment

Keep the targeted fixture-formatting changes: the defect is in test data generation, while production parsing is already culture-invariant. Introducing a JSON serializer or changing production parsing would add scope without addressing a broader need in this PR.

Files changed (2) +5 / -3

Bug fix (2) +5 / -3

KiroCrewParentResolverTests.csWrite sub-agent start times with invariant decimal formatting +2/-1

Write sub-agent start times with invariant decimal formatting

• Formats the epoch-seconds value in seeded sub-agent JSON with CultureInfo.InvariantCulture. This preserves a valid JSON number when tests run under a comma-decimal locale.

test/Capacitor.Cli.Core.Tests.Unit/Harness/Kiro/KiroCrewParentResolverTests.cs

KiroCrewSessionEndWatchTests.csWrite session metadata timestamps with invariant formatting +3/-2

Write session metadata timestamps with invariant formatting

• Uses invariant interpolation for session creation timestamps in two watcher tests. This prevents locale-specific time separators from changing their seeded metadata.

test/Capacitor.Cli.Core.Tests.Unit/Harness/Kiro/KiroCrewSessionEndWatchTests.cs

[realtonyyoung]

/agentic_review

[qodo-code-review]

Code Review by Qodo

🐞 Bugs (0) 📘 Rule violations (1) 🔗 Cross-repo conflicts (0) 📜 Skill insights (0)

1. The PR title breaks the commit naming rule 📘 Rule violation ⚙ Maintainability

Description

The PR title and first commit subject are `[AI-3386] Write the Kiro Crew test fixtures
culture-invariantly`, using a bracketed Linear identifier and omitting the required trailing GitHub
reference in (#number) form. Because merge-time commit validation checks this subject before
implementation review, the change is rejected even though the description contains `Closes #1250 —
AI-3386`.

Code

test/Capacitor.Cli.Core.Tests.Unit/Harness/Kiro/KiroCrewParentResolverTests.cs[1]

+using System.Globalization;

Relevance

● Weak

PR #659 accepted guidance allowing [AI-123] title prefixes while requiring references in the
description, contradicting this title-format finding.

PR-#659

ⓘ Recommendations generated based on similar findings in past PRs

Evidence

The supplied PR title and first commit subject are identical and begin with [AI-3386], confirming
that the Linear-style issue token is present in both user-facing and commit metadata. The title does
not end with the required (#digits) reference, while Closes #1250 — AI-3386 appears only in the
description; therefore the title and commit-subject requirements remain independently non-compliant.

Rule 2897961: Enforce single-clause imperative commit subject with GitHub issue reference and 80-char limit
Rule 2897991: PR description must contain both GitHub and Linear issue references; PR title must not contain issue IDs
Rule 2897978: Format PR titles as imperative commit subjects without explicit issue references

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
The PR title and primary commit subject use `[AI-3386]` and do not satisfy the required concise imperative format ending with a GitHub issue reference such as `(#1250)`. Remove the bracketed Linear identifier from the title and commit subject while retaining the description reference line `Closes #1250 — AI-3386`.

## Fix Focus Areas
- test/Capacitor.Cli.Core.Tests.Unit/Harness/Kiro/KiroCrewParentResolverTests.cs[1-1]
- test/Capacitor.Cli.Core.Tests.Unit/Harness/Kiro/KiroCrewSessionEndWatchTests.cs[1-1]
- test/Capacitor.Cli.Core.Tests.Unit.Harness/Kiro/KiroCrewParentResolverTests.cs[1-1]

## Recommended Fix
Rename the PR title and merge commit subject to one imperative clause of no more than 80 characters, remove `[AI-3386]`, and end the subject with the explicit GitHub issue reference `(#1250)`. Keep `Closes #1250 — AI-3386` together on the description's reference line.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

Dismiss ↗ | View ↗

---

Context sources

✅ Compliance rules (platform): 64 rules

✅ Cross-repo context — repo relationships

Review mode: 🚀 Fast: This is a small, localized test-fixture culture-invariance change with no production, security, API, or data-path impact.

Tip of the day

💡 Did you know, you can route each severity your way: inline, summary, both, or drop

More tips ↗ | Customize Qodo ↗ | Qodo docs ↗

[realtonyyoung]

Declining the title-format finding: this repo's merged history uses the [AI-N] Imperative subject form (e.g. #1232, #1216–#1219), GitHub appends (#N) at squash time, and the issue reference is in the description (Closes #1250 — AI-3386). Qodo's own relevance note cites PR #659 accepting this convention.