kube-coder preflight (run CI locally)

Mirrors .github/workflows/ci.yml so you catch failures before pushing instead of waiting on a CI round-trip.

Run kc-scope-pr first. This skill runs whole suites; that one tells you which suites your change actually reaches, and what it left untested. A green preflight over a suite that never touches your diff is not evidence. Run the subset named by $ARGUMENTS (default all). Report a concise PASS/FAIL summary per section at the end.

Run from the repo root (/home/dev/kube-coder).

Prereqs (install once, idempotent)

export PATH="$HOME/.local/bin:$PATH"
# helm (if missing)
command -v helm >/dev/null || {
  cd /tmp && curl -fsSL https://get.helm.sh/helm-v3.14.4-linux-amd64.tar.gz | tar xz \
    && install -m755 linux-amd64/helm ~/.local/bin/helm; cd -; }
# helm-unittest plugin (if missing)
helm plugin list 2>/dev/null | grep -q unittest \
  || helm plugin install https://github.com/helm-unittest/helm-unittest.git --version v0.5.2

helm — lint + template + unit tests

export PATH="$HOME/.local/bin:$PATH"
helm lint charts/workspace/ -f charts/workspace/tests/test-values.yaml
helm template test-ws charts/workspace/ -f charts/workspace/tests/test-values.yaml > /dev/null
helm unittest charts/workspace/
# controller chart too, if you touched it:
# helm lint charts/workspace-controller/ -f charts/workspace-controller/tests/test-values.yaml
# helm unittest charts/workspace-controller/

python — server.py unit + integration tests

make python-tests
# equivalently: cd charts/workspace && python3 -m unittest discover -s tests -p '*_test.py'

web — SPA typecheck, build, and vitest (BOTH apps)

There are two SPAs with separate lockfiles and CI jobs: the dashboard (charts/workspace/web) and the controller (charts/workspace-controller/web). CI runs each build (which is tsc --noEmit && vite build) then vitest — mirror both:

for dir in charts/workspace/web charts/workspace-controller/web; do
  yarn --cwd "$dir" install --frozen-lockfile   # first run only
  yarn --cwd "$dir" build      # tsc + vite
  yarn --cwd "$dir" test       # vitest run
done

Ignore the DOMException [NetworkError] … /oauth/terminal/ noise in TerminalPane tests — happy-dom tries to actually fetch iframe srcs; it's harmless and pre-existing. Only the final Test Files … / Tests … tally matters.

shell — validate the entrypoint scripts

start.sh ships via {{ tpl (.Files.Get "start.sh") . }}, so a syntax error there breaks every pod. Always:

bash -n charts/workspace/start.sh

For the SSH sidecar (embedded in YAML), rendering it through helm already covers syntax; helm template … | grep to eyeball the script if you edited it.

What CI runs that this does NOT

• Docker Build Smoke Test — a full image build (~2 min). Usually not worth reproducing locally; rely on CI. If you changed the Dockerfile, do build it.
• Trivy / Security Scanning — image + dependency CVE scans. CI-only.

Call those out in your summary so the user knows what's still pending after push.

Summary format

End with a table:

Section	Result
helm lint/template/unittest	✅ / ❌
server.py tests	✅ / ❌
dashboard SPA build + vitest	✅ / ❌
controller SPA build + vitest	✅ / ❌
shell bash -n	✅ / ❌

If everything passes, tell the user it's safe to run kc-ship-pr.