[eslint-refiner] ESLint Refiner — Daily Report 2026-07-13

[github-actions[bot]]

Overview

Today's review targeted no-github-request-interpolated-route — the newest rule added since the last recorded run that had no prior refinement issue. The rule is sound and already catches real anti-patterns in the live corpus; the two filed refinements close a grounded client-detection false negative and a diagnostic-precision gap.

The rule inventory has grown from 12 → 15 rules. Two of the three new rules (no-throw-plain-object, require-return-after-core-setfailed) were already reviewed in unrecorded runs (issues #45050, #45051) — those angles were deliberately skipped to avoid duplication.

Key metrics

Metric	Value
Rules in factory	15 (was 12 on 07-08)
Rule reviewed today	no-github-request-interpolated-route (first review)
Live true positives (rule working)	4 — route_slash_command.cjs:180/192/204/216
Live false positives	0
Issues filed	2
Duplicates explicitly skipped	2 (#45050, #45051)

Issues filed

1. Scope-resolution false negatives — global.getOctokit(...) is this repo's actual Octokit client factory (setup_globals.cjs:56; create_pull_request.cjs:85; create_agent_session.cjs:37) but is missing from the rule's getOctokit member-object allowlist {github, actions}. A const gh = global.getOctokit(token); gh.request(\GET ...${x}...`)` alias silently escapes. Also: the route argument is never scope-resolved even though the client is (asymmetry).
2. Diagnostic remedy mismatch — the single message always prescribes the {owner}/{repo} placeholder fix, which is inapplicable to opaque whole-route helpers (\${METHOD} ${routeVar}`) at add_reaction.cjs:193, add_reaction_and_edit_comment.cjs:266, add_workflow_run_comment.cjs:444`.

Analysis detail & deliberately-skipped findings

Rule works as designed. The value-into-path interpolations in route_slash_command.cjs (interpolating owner/repo/issueNumber directly into the path) are genuine true positives the rule correctly flags. No false positives were found across the non-test actions/setup/js/**/*.cjs corpus, which is disciplined.

Skipped — already owned by open issues:

• The ~22 JSON-RPC throw { code, message } sites in mcp_server_core.cjs / safe_outputs_handlers.cjs are covered by eslint-factory: no-throw-plain-object should offer an Object.assign(new Error(...)) suggestion #45050, which frames them as true-positives-to-convert (add an Object.assign(new Error(...)) suggestion). Filing an "exempt these as intentional protocol errors" issue would contradict that decision — deferred to the team's existing position.
• Aliased/destructured core.setFailed detection for require-return-after-core-setfailed is covered by eslint-factory: require-return-after-core-setfailed misses aliased/destructured core.setFailed #45051.

Ungrounded / not filed: name-only client match ignoring bindings; new Octokit() alias recognition (no such construction in corpus); this.github.request and .concat()-built routes (out of scope by design).

Process note: repo-memory was stale — it did not record the 07-09..07-12 runs that filed #45050/#45051. Reinforced the rule: always gh api dedupe before assuming a rule is unreviewed.

Next actions

• On the next run, diff index.ts imports against the 15-rule inventory for a potential 16th rule and lead with it.
• If no new rule appears, no-json-stringify-error and prefer-get-error-message are the remaining rules without a dedicated review issue.
• Verify fixes for the two filed issues keep static-literal const routes (update_project.cjs:855) unflagged.

References:

• §29228348376

Generated by 🤖 ESLint Refiner · 411.7 AIC · ⌖ 13 AIC · ⊞ 4.6K · ◷

• expires on Jul 13, 2026, 10:32 PM UTC-08:00