Importance: Medium | Status: Needs Follow-up (open QA + 1 real bug)

Mark task: Pending. The pulse itself ran end-to-end, but it surfaced an infra decision (how to run native/large suites) plus one dead lint config that should be fixed.

What was done

First run of the weekly repo test/lint pulse. READ-ONLY against repos: nothing committed, pushed, or modified; all installs/test runs in throwaway temp copies, cleaned up after. Source node_modules verified intact post-run.

• Discovered 16 git repos under Documents/Projects/Repos/ (dynamic .git scan). Non-git folders skipped, including sail-googleads per task.
• Detected test/lint config at root + depth-3 across all 16.
• Ran what the sandbox could: ESLint + Vitest on sail-command-center; node --test + eslint on the projects/sail-routines/google-ads routine.
• Wrote the report to Documents/Codex/repo-health-tests/2026-06-07.md with a grep-able aggregate line for the daily check's Section 8.

Results

Test/lint pass rate: 0 of 16 repos green (8 have no test/lint configured; of the 8 configured: 1 lint-only pass, 1 lint failure, 6 deferred/stub).

• sail-command-center — cleanest suite in the fleet (vitest, 8 specs, eslint). ESLint PASS. Tests BLOCKED: the repo's node_modules was installed on macOS, so the linux-arm64 rolldown/vitest native binding is missing. Not a code failure.
• projects/sail-routines/google-ads — declares test + lint but has no .test.js files (0/0) and no ESLint config, so lint errors out (exit 2). Dead lint command — real bug.
• sail-spanish-lp — real suite (vitest + tsc --noEmit) but no pre-installed deps; full install won't fit the sandbox. Deferred.
• sail-infrastructure / sail-litify — test/lint only in nested packages (workspace MCP servers; Salesforce LWC sfdx-lwc-jest). Heavy toolchains, deferred.
• 8 repos (google-workspace-cli, ha-law, sail-cases, sail-dashboard, sail-hr, sail-knowledge, sail-marketing-dashboard, sail-templates) have no test/lint at all — expected for content/metadata repos, valid finding.

Why the session ended

Completed the discovery pass. Two hard sandbox limits capped execution: (1) ~250 MB free on local disk + session volume 100% full, so a Next.js/Vite node_modules (300-500 MB) won't fit; (2) detached background installs don't survive between shell calls, and reused macOS deps fail for native-binding tools. These are infra constraints, not repo problems.

Recommended next actions

• Decide where native/large suites should run (see QA below). The cleanest fix is running this pulse on a runner with real local disk + matching arch, or pointing it at each repo's existing GitHub Actions.
• Fix the dead lint config in projects/sail-routines/google-ads (add an eslintrc or drop the script).