[eslint-refiner] ESLint Refiner — Daily Report 2026-07-22

[github-actions[bot]]

Overview

First refiner run since 2026-07-08. In that gap the eslint-factory plugin grew from 12 → 24 registered rules (~1 new rule/day), leaving 12 rules never reviewed. Today I reviewed the 4 newest (by index.ts import order) and filed 2 non-duplicate refinement issues. The newest rules are high quality and well-calibrated against the live corpus.

Key metrics

Metric	Value
Registered rules	24 (+4 shared utils)
Rules reviewed this run	4 (newest)
Rules reviewed ≥1× total	16 / 24
Never-reviewed backlog	8
Issues filed	2 (execFileSync gap; exitCode parity)
Live true-positives confirmed	3 (rules working)

Issues filed

1. Cover execFileSync in require-execsync-try-catch — the newest rule guards execSync (throws on non-zero exit) but not its identical-semantics sibling execFileSync; spawnSync (non-throwing) is covered by a different rule, so execFileSync is unguarded. Grounded: execFileSync runs in 4+ live scripts, manually try-wrapped in most (proving the invariant), but unwrapped in build_checkout_manifest.cjs:44/48. (HIGH, grounded)
2. no-core-error-then-process-exitcode forward-scan parity — it only inspects the adjacent statement, while its sibling no-core-error-then-process-exit scans forward; a benign intervening line (e.g. core.info(...)) between core.error() and process.exitCode = 1 silently defeats it. Since process.exitCode does not halt, intervening statements are common. (MED, soundness/parity)

Rule health — calibration checks

True-positives / true-negatives verified statically against non-test .cjs

• require-execsync-try-catch: correctly flags unwrapped start_mcp_gateway.cjs:824 execSync(\node "${p}"`); correctly **passes** wrapped sites (:482/:670/:693/:896, get_current_branch.cjs:18`). Well-calibrated.
• no-exec-interpolated-command: all live exec.exec / exec.getExecOutput use a static command string plus an args array (e.g. git_auth_helpers.cjs) → clean true-negatives; no false negative found.
• no-core-error-then-process-exit: correctly flags convert_gateway_config_shared.cjs:48-49 and :70-71 (core.error(); process.exit(1)).

Backlog observations (not filed this run)

• Cross-block continuation FN in the no-core-error-then-process-exit* family: they pair statements only within a single block, whereas the cousin require-return-after-core-setfailed already handles cross-block/loop/switch continuation. Candidate parity fix (ungrounded).
• process.exit(0) override: core.setFailed(m); process.exit(0) passes require-return-after-core-setfailed yet resets the exit code to success — a latent correctness footgun; niche/ungrounded (discussion-worthy, not an issue).
• no-core-exportvariable-non-string shares the non-string-kind util with no-core-setoutput-non-string, so it inherits the known gaps tracked for setOutput (only .length; misses unary/negative numeric literals, .number/.size). Review next; avoid duplicating prior setOutput issues.
• Minor dead code in require-execsync-try-catch:148-160 (withinDeferredBoundary computed then void-ed).

Next actions

• Continue draining the never-reviewed backlog (8 rules), prioritizing the thorough require-return-after-core-setfailed, no-github-request-interpolated-route, and no-throw-plain-object.
• On landing: re-verify build_checkout_manifest.cjs:44/48 flagged and wrapped execFileSync sites clean; confirm the exitCode forward-scan leaves existing valid cases green.
• Strategy + findings persisted to repo-memory (eslint-refiner branch) for continuity.

References: §29895687838

Generated by 🤖 ESLint Refiner · age00 312 AIC · ⌖ 12.9 AIC · ⊞ 4.6K · ◷

• expires on Jul 22, 2026, 10:18 PM UTC-08:00

[github-actions[bot]]

This discussion has been marked as outdated by ESLint Refiner.

A newer discussion is available at Discussion #47497.