kandev/docs/plans/kubernetes-executor-runtime-disclosure/task-03-desktop-mobile-e2e.md

id, title, status, wave, depends_on, plan, spec

id	title	status	wave	depends_on	plan	spec
03-desktop-mobile-e2e	Desktop and mobile runtime E2E	complete	3	02-frontend-runtime-disclosure	plan.md	../../specs/kubernetes-executor/spec.md

Task 03: Desktop and mobile runtime E2E

• Acceptance: the existing real kubeconfig Kind case validates the exact Pod details, status, cube glyph, Refresh, and settings navigation inside the desktop task disclosure without increasing the 12-case lifecycle catalog.
• Acceptance: one isolated mobile-chrome case taps a named Kubernetes executor control and verifies Drawer parity, 44 px target size, and no document horizontal overflow without requiring a live cluster.
• Acceptance: the E2E-owned Kind cluster/image/process teardown remains exact; the user-requested seed lab and the main Kandev instance on port 9998 are not stopped or modified.
• Verification: cd apps && pnpm install --frozen-lockfile && cd web && E2E_FAIL_ON_FLAKY=1 pnpm e2e:run --host --project mobile-chrome tests/settings/mobile-kubernetes-task-environment.spec.ts -- --retries=0 && E2E_FAIL_ON_FLAKY=1 KANDEV_E2E_KIND_BIN=/tmp/kandev-task06-pinned-tools/kind KANDEV_E2E_KUBECTL_BIN=/tmp/kandev-task06-pinned-tools/kubectl KANDEV_E2E_CONTAINERS=1 pnpm e2e:run --host --project containers tests/kubernetes -- --grep 'launches through kubeconfig with Pod exec and a loopback-only agentctl forward' --retries=0 && cd ../.. && git diff --check
• Files likely touched: apps/web/e2e/tests/kubernetes/kubernetes-executor.spec.ts and apps/web/e2e/tests/settings/mobile-kubernetes-task-environment.spec.ts.
• Dependencies: Task 02 completed frontend behavior.
• Parallelism: sequential.
• Inputs: spec scenarios; plan E2E section; existing Kubernetes fixture ownership and mobile overflow helpers.
• Output contract: report exact discovery/pass counts, timings, artifact paths, source-freshness evidence, teardown audit, files changed, blockers/risks, and synchronize task/plan status.

Results

• RED: the first isolated mobile run reached the Drawer but showed the generic no-environment state and Reset action because the intentionally unreachable cluster failed before creating a persisted task environment. The fixture was corrected to seed only its isolated worker database and intercept only the exact filtered read-only Kubernetes status request.
• GREEN: the fresh mobile-chrome command discovered exactly one test and passed 1/1 on the first attempt in 4.3 seconds overall (2.5 second test body), including Drawer parity, the 44 px trigger, canonical settings navigation, absence of Reset, and no horizontal overflow.
• GREEN: the fresh pinned Kind command extended the existing kubeconfig case without changing the 12-case catalog, discovered exactly one focused test, and passed 1/1 on the first attempt in 1.2 minutes overall (7.4 second test body). It verified the cube glyph, exact live Pod details, Refresh, no generic empty-resource text, and canonical settings navigation.
• Final teardown found only the separately requested seed Kind cluster, image, and container. The E2E-owned cluster, images, processes, and ownership markers were absent; the seed lab and main Kandev instance on port 9998 were not modified.