added-test-summary

Report the tests a branch adds (vs its base), run only those, and format a paste-ready summary. Count only genuinely-new tests — new test files, plus new cases inside modified files; never pre-existing ones.

1. Base

The branch's PR base, else main:

BASE=origin/$(gh pr view --json baseRefName -q .baseRefName 2>/dev/null || echo main)
git fetch "${BASE#origin/}"

2. Inventory added tests

# new test files — every case in these is new (.test.* = unit, .spec.* = e2e)
git diff --name-status --diff-filter=A "$BASE"...HEAD | grep -iE '\.(test|spec)\.(js|ts|jsx|tsx)$'
# modified test files — only some cases may be new
git diff --name-status --diff-filter=M "$BASE"...HEAD | grep -iE '\.(test|spec)\.(js|ts|jsx|tsx)$'
# new cases within a modified file
git diff "$BASE"...HEAD -- <file> | grep -E '^\+\s*(it|test)(\.(only|skip|each))?\('

Count only it/test. A describe/test.describe is a suite wrapper, not a test — count the cases inside it, not the wrapper.

3. Run only those

Pick the runner from the package's own test script (check it; don't assume). Run under the repo's Node (.node-version) against local Postgres:

• vitest — every package except mobile and e2e-tests: npx vitest run <file> [-t "<substring>"]
• jest — @tamanu/mobile: npx jest <file> [-t "<substring>"] --verbose
• Playwright e2e — @tamanu/e2e-tests (*.spec.ts): npx playwright test <file> [-g "<title>"]

Always pass the specific file — never a bare workspace script (npm test, npm run e2e-test), which runs the whole suite. Run a new file whole; use -t/-g to scope a modified file to its new cases.

When a runner can't run locally, cite the branch's CI run (gh run list --branch <branch> → gh run view --log) instead of reporting a false failure:

• stale node_modules (e.g. an es-toolkit/compat / defaultsDeep error) → the runner won't start;
• e2e needs a running stack — a reachable facility frontend + packages/e2e-tests/.env (FACILITY_FRONTEND_URL, TEST_EMAIL, TEST_PASSWORD). Without it, list the added specs/cases from the diff and pull pass/fail from CI.

4. Report

Head with the branch + commit SHA. Group by runner → file, one line per added case (✓, or the CI status when cited). Don't collapse to per-file counts — show each test name. End with per-runner and grand totals of added tests; keep pre-existing/skipped cases out of the totals (mention them only to explain a -t/-g filter), and mark any runner cited from CI.