Running tests

Every Ionize repo uses Vitest as its primary test runner (ADR 0020). Most repos also have Playwright for end-to-end.

Per repo

dashboard

cd dashboard
bun run test          # vitest run
bun run test:watch    # vitest watch
bun run test:e2e      # playwright test

Tests cover:

services/kratos/* adapters (mocked Ory clients).
lib/* utilities.
Route handlers (with mocked services).
A small Playwright suite that exercises login → identity create → identity delete.

auth

cd auth
bun run test
bun run test:e2e

Tests cover:

Captcha + breach-check integration.
Flow rendering with mocked Kratos.
Playwright login flow.

platform

cd platform
bun run test

Tests cover:

The reload sidecar.
init-db.sql shape verification.
Compose YAML validation.

sdk

cd sdk
bun run test

Tests cover:

Encryption format round-trip.
HKDF derivation.
Brute-force counter logic.
Settings cache TTL.

canvas

cd canvas
bun run test

Tests cover component behavior (Vitest + React Testing Library) and visual regression via Playwright for a small set of components.

site

cd site
bun run test:e2e

Site is mostly content; Vitest unit tests are minimal. The Playwright suite exercises the OAuth2 playground.

ionctl

cd ionctl
bun run test

Tests cover CLI command parsing and the bundled stack file existence.

deploy

cd deploy
bun run test                     # frontend tests (Vitest)
cargo test --manifest-path src-tauri/Cargo.toml  # backend tests (Rust)

Running everything

From the parent Ionize/ directory:

for repo in dashboard auth platform sdk canvas site ionctl deploy; do
  echo "=== $repo ==="
  cd $repo
  bun run test || echo "FAIL: $repo"
  cd ..
done

CI

Every push runs the test suite for the affected repo. CI workflows live at .github/workflows/ci.yml in each repo.

When tests fail in CI but pass locally

Common causes:

Environment vars. Some tests pull from process.env. CI may not have them set.
Snapshot files. Vitest's toMatchSnapshot writes to disk; if CI's snapshot differs from local, you might have stale snapshots committed.
Bun version mismatch. Local Bun vs CI's Bun may behave differently. Pin via package.json's engines.
Network access. CI runners may not have outbound to the same external services your local tests touch. Mock those.

Coverage

bun run test -- --coverage

Coverage isn't enforced as a CI gate by default, per directive (No Coverage Until User Says Push), iteration doesn't fail builds on coverage thresholds.

On this page