[quality] handlePullReq 'None found' setFailed arm is unreachable in dist/ — pullRequestHandlers is a static non-empty const #450

Finding

src/pullReq/handlePullReq.ts:44-48 still carries the pre-plugin-registry failure arm:

if (runConfig.length === 0) {
  if (pullRequestHandlers.length === 0) {
    core.setFailed('please provide a list of space delimited commands / jobs to run. None found')
  }
  return
}

pullRequestHandlers is declared six lines above (handlePullReq.ts:18) as a const array of seven handlers (requireMatchingLabel, ownersLabel, blunderbuss, lgtmOnPullRequest, approveOnPullRequest, okToTestOnPullRequest, tideOnPullRequest). Nothing in src/ mutates it, so pullRequestHandlers.length === 0 can never be true in the shipped dist/index.js and the setFailed call is dead.

It only counts as covered because the unit suite empties the registry to reach it:

• __tests__/pullReqTest/handlePullReq.test.ts — snapshots registeredHandlers = [...pullRequestHandlers], clears the array in setup, and asserts setFailed is called exactly once with the None found message.
• __tests__/pullReqTest/readOnlyFork.test.ts:73-78 — same length = 0 / restore pattern.

Evidence (main @ 511bd12, local runs 2026-10-11):

• npx vitest run --coverage (151 files / 2089 tests): 100 % statements, handlePullReq.ts fully covered.
• npm run test:coverage:e2e (72 files / 452 tests, bundle run against dist/index.js): handlePullReq.ts line 46 is the only uncovered statement in the file.
• Line-level cross-reference of the two coverage-final.json files: this is one of a handful of residual e2e-only lines; the others are test-only reset*() helpers or ground already held by refactor(auth): drop the 14 unreachable commenter-auth catch wrappers #449.

This is the same finding class as #366 (fixed by #421): production code that nothing in the action can execute, kept green by a unit test that exists only to exercise it. The generic handlers.length === 0 debug no-op in src/utils/events.ts:48 is a different case — runEventHandlers is shared by four registries and that guard is reasonable to keep.

Recommendation

One PR:

• Drop the inner if (pullRequestHandlers.length === 0) { core.setFailed(...) } from handlePullReq.ts, leaving if (runConfig.length === 0) return (or fold it into the switch below).
• Remove the None found assertion test in __tests__/pullReqTest/handlePullReq.test.ts and the pullRequestHandlers.length = 0 scaffolding it needed; keep the "does not fail when no jobs are configured but a handler is registered" case.
• Check whether readOnlyFork.test.ts:73-78 still needs to clear the registry once the arm is gone (it may — it isolates the fork short-circuit from the real handlers), and leave it if so.
• npm run pack so dist/ matches.

Priority

• Impact: low (dead code; no behaviour change, slightly misleading unit coverage)
• Effort: low

Filed by quality agent (hold-gated mode)

🐝 Hive Agent: quality | Instance: hosted-available-lke648397-260827-5q9t | SHA: 511bd12

— hive: agent=quality backend=copilot model=claude-fable-5.1 copilot=1.0.88