[dinwwwh]

🤖 Generated with Claude Code

https://claude.ai/code/session_01VVaw7RYWCsn8myBE4Rncuq

What

Test-only. Stryker on main after #363 reports 77 survivors, the same count as before #363. Two survivors are new:

• const others = Array.isArray(anyOf) ? nonNullBranches(anyOf) : []: [] replaced by ["Stryker was here"]. This is the existing equivalent survivor on that line, renamed by fix(downgrader): follow a null pair nested in a null pair around a $ref loop #363.
• The cycle guard's return [] in nonNullBranches, replaced by ["Stryker was here"]. This one is real. The fix(downgrader): follow a null pair nested in a null pair around a $ref loop #363 cycle test passes either way, because the only branch left is the revisited one, so a filler branch and no branch both end the walk.

The new assert adds a $ref beside the cycle (B = { anyOf: [{ anyOf: [B, null, $ref A] }, null] }, with A a pair around $ref B). The guard leaves only the $ref, so the walk follows it back to B and keeps the pair. A filler branch would make it give up and drop the null. Replacing the guard's result by hand now fails the test.

---

Generated by Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[codspeed]

Merging this PR will not alter performance

✅ 29 untouched benchmarks
⏩ 2 skipped benchmarks¹

---

_{Comparing claude/project-thread-rixcs1 (b3e2536) with main (69c521c)}

Footnotes

1. 2 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[pullfrog]

✅ No new issues found.

Reviewed changes

• Test: follow a $ref beside an object cycle — in the nullable-pair $ref-loop test, after the existing cycle assertion collapses A to a bare $ref B, the change pushes ref('A') beside the cycle and asserts A keeps its nullable pair { anyOf: [ref('B'), onlyNull] }, killing the nonNullBranches cycle-guard survivor (return []).

Test-only change; verified the full suite is green and that the new assertion fails when the guard's return [] is replaced with a filler branch, so it kills the intended mutant.

  ^{｜ View workflow run ｜ Using deepseek-v4.1-flash (free via Pullfrog for OSS) ｜ 𝕏}