[quality] validate-metrics.mjs optional-section validation branches have no unit coverage

Start with tests/validate-metrics.test.mjs and the runScriptWithFixtures helper in tests/helpers.mjs, then inspect the corresponding optional-section branches in scripts/validate-metrics.mjs. Add fixture-driven cases for the fourteen listed outcomes, asserting each exit status and structured validator error, and run node --test --experimental-test-coverage to verify the suite passes.

Finding

scripts/validate-metrics.mjs reports 100% lines but 30.00% branches under node --test --experimental-test-coverage. Reading tests/validate-metrics.test.mjs against the script confirms that its optional-section validation paths are never exercised. Fourteen distinct validation outcomes had no test:

• metric with an absent id
• metric with a null value
• empty metrics array
• omitted entry missing id
• lifecycle card with a non-finite value
• lifecycle omission missing reason
• fully-populated valid lifecycle section
• time series missing sourceUrl
• series point with a non-finite value
• fully-populated valid series section
• breakdown whose values is not an array
• breakdown missing its label
• non-finite breakdown value
• fully-populated valid breakdown section

These are the branches that only run when a metrics file carries the optional referenceArchitectureLifecycle, series, or breakdown sections. data/metrics.json is a generated artifact, so a partially-written or malformed regeneration is exactly the case these guards exist for — and exactly the case nothing currently tests.

Coverage evidence

• Unit: node --test --experimental-test-coverage (node v26.8.2), run locally at 00b44df (current main). 55 tests pass. scripts/validate-metrics.mjs: 100.00% lines, 30.00% branches, 100% funcs.
• End-to-end: the repository has no end-to-end or browser suite — no playwright/cypress/e2e dependency or config exists and no workflow runs one — so no end-to-end coverage can be claimed either way for this module. Tracked separately, not asserted here.
• Measurement limitation: the two sources cannot be combined. The reported branch percentage will not move when these tests are added, because tests/helpers.mjs runScriptWithFixtures executes a temp-dir copy of the script in a subprocess, which node --test instrumentation does not attribute back to the repo path. Evidence of added coverage is therefore behavioral: each test asserts the specific exit status and the specific validator error message for its branch.

Per the coverage priority rules this is priority 2 / medium at most, and it is scoped to unit coverage only.

Recommendation

Add fixture-driven tests for each of the fourteen outcomes above, asserting exit status and the matching structured error text rather than a coverage number.

Priority

• Impact: medium — malformed generated metrics can bypass the validator's optional-section guards untested
• Effort: low — test-only, uses the existing runScriptWithFixtures helper

Filed by quality agent (hold-gated mode)

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