normalizeAmount takes an _accountType parameter it never reads

[KenTaniguchi-R]

Mechanism

pnpm lint exits 1 with 1,024 errors. 981 of the 982 reported files are under .stryker-tmp/, the mutation-testing sandbox.

.stryker-tmp/ is listed in .gitignore:53, but ESLint 9 flat config does not read .gitignore. eslint.config.mjs:15 calls globalIgnores([...]) with only the eslint-config-next defaults (.next/**, out/**, build/**, next-env.d.ts) — .stryker-tmp is not among them.

Impact

Any developer who has ever run pnpm test:mutate has a permanently red pnpm lint locally. CI passes only because the runner starts from a clean checkout, so the failure is invisible where it would be caught and constant where it is most annoying. It also buries the one genuine finding.

Reproduction

$ pnpm test:mutate:incremental   # creates .stryker-tmp/
$ pnpm lint
...
1037 problems (1024 errors, 13 warnings)

Attribution by top-level directory:

981  .stryker-tmp
  1  src

Fix

Add .stryker-tmp/** to globalIgnores in eslint.config.mjs, or wire .gitignore in via includeIgnoreFile from @eslint/compat.

The single real source finding is src/lib/money.ts:28 — normalizeAmount(amountCents, _accountType) takes an account-type parameter it never reads. Worth deleting the parameter outright: its presence implies account type affects normalization when it does not, which is misleading next to the sign-convention problems in #84.

Found in UI audit 2026-08-29.

[KenTaniguchi-R]

Half of this has already landed — rescoping

The ESLint half is fixed on main. eslint.config.mjs now carries .stryker-tmp/** in globalIgnores, with a comment explaining why:

// Stryker leaves sandbox copies of the whole tree behind. Linting those
// repeats every finding once per sandbox and fails the run on code that
// is not ours to fix.
".stryker-tmp/**",

So the headline — "pnpm lint is red for anyone who has run mutation testing" — is no longer true. Verified on a clean tree with a populated .stryker-tmp/: pnpm lint reports 1 problem (0 errors, 1 warning), down from 1,037.

What actually survives

The single real source finding from the original report:

src/lib/money.ts
  30:54  warning  '_accountType' is defined but never used  @typescript-eslint/no-unused-vars

normalizeAmount(amountCents, _accountType) takes an account-type parameter it never reads. Worth deleting outright: its presence implies account type affects normalization when it does not, which is actively misleading next to the sign-convention work in #84/#95.

Blast radius is small — 2 production callers (src/app/api/import/route.ts:186, src/lib/plaid/sync.ts:181) and 2 test files, where the parameter is threaded through ~12 assertions and 2 property tests that exist only to prove account type doesn't matter.

Retitling to match what's left. PR incoming.