[eslint-refiner] ESLint Refiner daily report - 2026-10-16

[github-actions[bot]]

Overview

Mined eslint-factory rule history for false positives/negatives not yet addressed. Two fresh, non-duplicate gaps were confirmed by reading rule source and test fixtures directly, and two other candidates were ruled out as duplicates of prior work before filing anything.

Key metrics

• Rules/utilities reviewed in depth: 9 (try-catch-rule-utils.ts, require-escaped-regexp-interpolation.ts, no-unsafe-catch-error-property.ts, non-string-kind.ts, require-parseInt-radix.ts, command-initializer-utils.ts, no-duplicate-constant-values.ts, no-async-foreach-callback.ts, plus its test fixture)
• Issues created: 2
• Candidates dropped as duplicates/already-decided: 2

Details

Issue 1: resolveInitializer never resolves module-scope command/route constants

command-initializer-utils.ts (lines 94-101) refuses to resolve module/global-scope const bindings even though it already has a write-once safety check that would make this safe -- the exclusion is purely scope-based, not safety-based. This utility backs 5 rules: no-exec-interpolated-command, no-child-process-interpolated-command, no-github-request-interpolated-route, require-error-code-for-github-api-throw, require-error-code-in-thrown-error. A module-level const built from a template literal (e.g. a shared git command string) and later passed to exec.exec() silently bypasses all five.

Issue 2: no-async-foreach-callback misses computed access and named async references

The rule own test fixture lists two forms as valid (unflagged) that are the exact bug class it exists to catch: items["forEach"](async x => {}) (computed-property access, skipped by an unconditional callee.computed early return) and items.forEach(handler) where handler is a named async function (never resolved back to its declaration). Both are concrete, reproducible gaps with no scope-analysis prerequisite for the first.

Candidates ruled out as duplicates

• require-parseInt-radix: aliased/destructured parseInt bindings (const p = parseInt, const { parseInt } = Number) are unflagged -- but this is explicitly decided out of scope, per both the current source comment and issue eslint-factory: require-parseInt-radix misses computed/indirect parseInt access #42017 acceptance criteria. Not re-filed.
• no-duplicate-constant-values: empty-string values fall through to the generic min-group-size-2 bucket instead of the size-3 guard given to number/boolean/null -- this duplicates issue no-duplicate-constant-values: string duplicates use a flat min-group-size of 2, unlike the size-3 guard given to numbers/boolean #59647, which auto-expired unfixed. The underlying gap is still real but re-filing identically would not add value; flagged in memory for a future run with better grounding.

Next actions

• Track the two new issues through to landing; re-grep the affected files afterward to confirm acceptance criteria were met.
• A future run should revisit no-duplicate-constant-values (no-duplicate-constant-values: string duplicates use a flat min-group-size of 2, unlike the size-3 guard given to numbers/boolean #59647) with concrete grounding (an actual duplicated empty-string pair in actions/setup/js) rather than re-filing the same latent-risk framing that expired.
• Rules not yet audited this pass: no-json-stringify-equality, no-math-minmax-array-spread, prefer-number-isnan, and the require-nan-check-after-* family.

Generated by 🤖 ESLint Refiner · claude · agent · 161.5 AIC · ⌖ 0.878 AIC · ⊞ 5.3K · ◷

• expires on Oct 16, 2026, 7:44 PM UTC-08:00