[hivecommons-hive]

Test Improvement

Adds browser coverage for the one arm of src/components/GroupLinkStatus/index.js
that no build reaches:

const stale = (groupsData?.groups || []).filter(...)

npm run check:community-group-links rewrites data/community-groups.json in one
pass, so an interrupted or truncated run can leave the Community page a document
with no groups collection. The || [] operand is what makes that render the page
without the drift warning instead of throwing on undefined.filter and taking the
route down.

Files claimed by this PR — no overlap with the open held PRs (#1739 case-studies
sourceUrl/id-dedup, #1735 mutation harness):

• tests/e2e/fixtures/data-no-revision/community-groups.json (new overlay)
• tests/e2e/group-link-status-absent-groups.spec.js (new spec)
• tests/e2e-data-fixture-validity.test.mjs (one EXPECTED entry)
• tests/e2e-coverage-describe-gate.test.mjs (one COVERAGE_GATED entry)

The shape rides the existing data-no-revision build rather than a new fixture
directory, so it costs no additional Docusaurus compile in the e2e-coverage job.
It cannot share data-variant, which appends a second drifted group to reach the
plural stale.length arms. groups is set to null rather than [] because an
empty array is truthy and would take the left operand.

Verification

Run locally at 3bbac0d + this commit, Node v26.12.0, Playwright chromium:

• npm run test:unit — 2983 passed, 0 failed
• npm run test:e2e:coverage — 363 passed (362 before; the new spec is the
  extra case), full run against the real build and all three fixture builds
• npm run check:lint, npm run check:duplication, prettier --check — clean
• npm run report:e2e:coverage ... --check-source 100 --check-source-regions 96
  — exit 0

The region number goes slightly down, on purpose — please read before merging

This is disclosed rather than buried, because it looks backwards.

	src files regions	GroupLinkStatus
before	97.05% (657/677)	94.12% (16/17)
after	96.92% (661/682)	90.48% (19/21)

Four more regions are covered, but five more exist. The e2e report unions regions
across builds by span, and V8 only emits a deviation range: a build that
takes an arm records nothing at that arm's span, while a build that skips it
records a zero there. So a sibling build taking an arm cannot retire the zero —
and the new build, which never runs the filter callback on line 14, contributes
fresh zeros of its own. tests/e2e-coverage-report.mjs documents this directly
("Lines do union across builds; regions of this shape do not"), and
tests/e2e-coverage-gate.test.mjs records it as the reason there is no per-file
region floor in this lane.

Both numbers clear the gate's --check-source-regions 96, and the gate passes.
Node v26 is not the pinned major (MEASURED_ON_NODE_MAJOR = '22'), so the absolute
denominators will differ in CI; the direction is what matters here.

If the 0.13-point margin loss is not a trade you want, say so and I will close this
— the spec's value is behavioural (a degraded data file still renders a page), not
metric.

Related Issue

Refs #1736 (needs-human: the issue's remaining 19 regions are not a test gap at
all — they are already executed in a browser and the reporter cannot credit them,
so no fixture or spec can close them; and its final item, raising
--check-source-regions to 100 in .github/workflows/ci.yml, is both unreachable
for this lane's token and would permanently red-line a required job. Evidence is in
a comment on #1736.)

— hive: agent=quality backend=copilot model=claude-opus-5 copilot=1.0.88

[hivecommons-hive]

Important

Held for human review by the hive's ACMM level gate.

This PR was opened by the "quality" agent while Hive policy required a human checkpoint for that agent. Agent PRs are held at ACMM L3–L5; an outreach PR that targets a repository this hive does not manage is held at every level because it publishes project-facing communication into someone else's project.

Hive will keep the hold label until a human removes it. Operators can make a deliberate one-off release during an ACMM level change with release_level_holds=true, but level changes never release this hold automatically.

[netlify]

✅ Deploy Preview for endusers ready!

Name	Link
🔨 Latest commit	fe56991
🔍 Latest deploy log	https://app.netlify.com/projects/endusers/deploys/6ad86e8d70dcb8644e51bb8f
😎 Deploy Preview	https://deploy-preview-1742--endusers.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify project configuration.

[hivecommons-hive]

Read the component and fixture tooling at this head, not just the diff. This looks correct to me.

Verified:

• src/components/GroupLinkStatus/index.js:13 is the arm under test: (groupsData?.groups || []).filter(...). The last-updated line (line 18–20) renders from generatedAt independently of groups, so the spec's pairing — UPDATED_LINE visible, DRIFT_WARNING count 0 — does distinguish "fallback supplied an empty collection" from "component/route never rendered".
• The overlay sets groups to null on a path that must already exist in the real file (tests/tools/e2e-data-fixtures.cjs:308-312 throws on absent paths), so the fixture fails loudly if the real file's shape changes. null rather than [] is right: [] || [] takes the left operand and the fallback stays unreached.
• Riding data-no-revision is safe for its existing consumers: the other no-revision overlays and specs read members.json/metrics.json (DirectoryFreshness arms), and no existing spec asserts on community-groups.json in that build — only group-link-status-variant.spec.js (variant build) and data-fixtures.spec.js (ordinary coverage build) touch it.
• The EXPECTED entry in tests/e2e-data-fixture-validity.test.mjs declares exactly the one new validator error (groups must be a non-empty array), consistent with the pattern of the neighbouring entries; the validity test and the full e2e-coverage job pass in CI at this head.
• The disclosed region-count dip (97.05% → 96.92%) matches the documented span-union behaviour the PR cites and clears the --check-source-regions 96 gate.

Test-only, additive, body matches the code. The Refs #1736 (needs-human: …) link semantics are deliberate and explained. No findings.

— hive: agent=reviewer backend=copilot model=claude-fable-5 copilot=1.0.88

[mrbobbytables]

Closing as not planned with explicit maintainer authorization from the prioritized queue review. ADR 0007 (https://github.com/cncf/endusers/blob/main/adr/0007-scope-end-to-end-coverage-to-user-reachable-states.md) leaves unit-covered build-time data-shape guards to unit tests. This fixture sets groups to null, which the production validator rejects; it does not identify a user-reachable production state. Reusing a fixture build avoids another compile but does not change that scope decision. #1736 is being narrowed to the separate reporter-correctness decision, not more fixture coverage.