Confirmed via same-session head-to-head blind adjudication: TrueMend 2.1% wrong-match vs. Deno lint's 5.0% — no new remediation needed, since the fixes from the Biome campaign already carried over to this corpus.
Deno lint is the built-in linter shipped with the Deno runtime — a third,
independent TypeScript/JS comparison point on the same next.js corpus we
used for ESLint and Biome.
No separate remediation campaign — and that's the point
The collector (collect_denolint) was already built alongside Biome's;
noExplicitAny, no-window, and no-window-prefix are excluded as
Deno-runtime-portability nags that don't apply to a browser-facing Next.js
codebase. Once built, this benchmark needed no detector fixes of its own —
the Round 3 remediation from the Biome campaign applies to the whole
TypeScript/JS surface, not just one competitor's ruleset, so it carried over
automatically.
Result
TrueMend: 2.1% wrong-match. Deno lint: 5.0%. 338 TrueMend findings and 140 Deno-lint findings, judged blind, same-session head-to-head. Worth-fixing came back essentially tied — 2.4% vs. 2.1% — within noise at this sample size, so we're not claiming an edge there either way.
The residual, not hidden
Seven findings remain wrong across six categories, all long-tail. Two are
the same cq-deeply-nested-callbacks trade-off disclosed in the Biome
write-up — kept active on purpose, for recall, not an oversight.