Replace Tart VM infrastructure with direct macOS host testing

.github/workflows/test.yml

@@ -158,6 +158,24 @@ jobs:
         if: always()
         run: kill $(cat /tmp/mock-pid) 2>/dev/null || true
 
+  macos-e2e:
+    name: macos e2e (L5)
+    runs-on: macos-latest
+    # Only on release tags or manual dispatch — these are slow and destructive.
+    if: startsWith(github.ref, 'refs/tags/v') || github.event_name == 'workflow_dispatch'
+    timeout-minutes: 45
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+
+      - name: Set up Go
+        uses: actions/setup-go@v5
+        with:
+          go-version-file: "go.mod"
+
+      - name: Run macOS E2E (release tier)
+        run: make test-vm-release
+
   cli-compat:
     name: old-cli compat
     runs-on: macos-latest

CLAUDE.md

@@ -21,7 +21,7 @@ make build-release VERSION=0.25.0    # optimized + UPX
 make test-unit                       # L1 (~15s) — pre-push hook
 make test-integration                # L2 (~75s) — real brew/git/npm in temp dirs
 make test-e2e                        # L4 compiled binary
-make test-vm-release                 # L5 VM (~20m) — before tagging
+make test-vm-release                 # L5 destructive macOS (~20m) — before tagging
 make test-destructive                # L6 — actually installs
 make test-coverage                   # coverage.out + coverage.html
 
@@ -60,7 +60,7 @@ internal/
   ui/                # bubbletea Model pattern, lipgloss styling
   updater/           # Auto-update: check GitHub → download → replace
 test/{integration,e2e}/   # //go:build integration | e2e (+ vm, destructive, smoke)
-testutil/            # shared helpers + Tart VM helpers
+testutil/            # shared helpers + MacHost (destructive E2E on real macOS)
 scripts/
   install.sh         # curl|bash installer
   hooks/             # pre-commit, pre-push (install via `make install-hooks`)

CONTRIBUTING.md

@@ -38,14 +38,14 @@ Tests are split across six tiers. Which one runs where:
 | **L2 Integration** | Real `brew` / `git` / `npm` against temp dirs; real `httptest` servers | `make test-integration` (~75s) | CI on push/PR |
 | **L3 Contract schema** | JSON schema validation against [openboot-contract](https://github.com/openbootdotdev/openboot-contract) | (runs in CI only) | CI on push/PR |
 | **L4 E2E binary** | Compiled binary driven by scripts; `-tags=e2e` | `make test-e2e` | CI on release |
-| **L5 E2E VM** | [Tart](https://github.com/cirruslabs/tart) macOS VMs (install Homebrew, run real flows) | `make test-vm-quick` (2 min) / `test-vm-release` (20 min) / `test-vm-full` (60 min) | Manual, before tagging a release |
+| **L5 Destructive macOS** | Runs against a real macOS host (installs packages, modifies `~/.zshrc`, writes `defaults`) | `make test-vm-quick` / `test-vm-release` / `test-vm-full` — requires `CI=true` or `OPENBOOT_E2E_DESTRUCTIVE=1` | GH Actions `macos-latest` on release tags + manual dispatch |
 | **L6 Destructive** | Actually installs real packages into a real system | `make test-destructive` / `test-smoke` | CI on release, plus manual `workflow_dispatch` |
 
 Rules of thumb:
 
 - **Local dev:** run nothing manually if hooks are installed. `make test-unit` on demand when you want a sanity check. Skip L2+ unless you're touching code that interacts with real brew/git/npm.
 - **Before pushing:** `make test-unit` (the pre-push hook does this automatically).
-- **Before tagging a release:** `make test-vm-release` locally (needs Tart).
+- **Before tagging a release:** trigger the `macos-e2e` job via GitHub Actions (manual dispatch or tag push). To run locally on a throwaway macOS machine: `OPENBOOT_E2E_DESTRUCTIVE=1 make test-vm-release`.
 
 ## Git Hooks
 

Makefile

@@ -44,31 +44,35 @@ test-all:
 	$(MAKE) test-coverage
 
 # =============================================================================
-# VM-based E2E tests (Tart VMs) — three levels
+# Destructive macOS E2E tests — three levels
 # =============================================================================
+#
+# These tests install real packages and modify ~/.zshrc / macOS defaults on
+# the host they run on. They are intended for ephemeral macOS CI runners
+# (GitHub Actions macos-latest) or a throwaway VM.
+#
+# On a developer machine `go test -tags="e2e,vm"` will skip unless you set
+# OPENBOOT_E2E_DESTRUCTIVE=1 (see testutil/machost.go). Don't set that
+# unless you mean it.
 
-# L1: Quick validation (~2min) — run after code changes
-#     Runs TestVM_Infra only: boots a VM and checks SSH/arch/tools, no package installs
+# L1: Quick sanity (~1min) — host/arch checks only, no package installs
 test-vm-quick: build
 	go test -v -timeout 5m -tags="e2e,vm" -run "TestVM_Infra" ./test/e2e/...
 
-# L2: Release validation (~20min) — run before tagging a release
-#     Core user journeys: dry-run safety, install + verify, diff/clean cycle,
-#     manual uninstall recovery, full setup, error messages
+# L2: Release validation (~20min) — core user journeys
 test-vm-release: build
 	go test -v -timeout 30m -tags="e2e,vm" \
-	  -run "TestVM_Infra|TestVM_Journey_DryRun|TestVM_Journey_FirstTimeUser|TestVM_Journey_ManualUninstall|TestVM_Journey_DiffConsistency|TestVM_Journey_FullSetup|TestVM_Journey_ErrorMessages" \
+	  -run "TestVM_Infra|TestVM_Journey_DryRunIsCompletelySafe|TestVM_Journey_FirstTimeUser|TestVM_Journey_FullSetupConfiguresEverything|TestE2E_DryRunMinimal|TestE2E_SnapshotCapture" \
 	  ./test/e2e/...
 
-# L3: Full validation (~60min) — run for major releases or CI
-#     All 48 tests: journeys + edge cases + commands + interactive
+# L3: Full validation (~60min) — everything under -tags="e2e,vm"
 test-vm-full: build
 	go test -v -timeout 90m -tags="e2e,vm" ./test/e2e/...
 
 # Aliases
 test-vm: test-vm-release
 
-# Single VM test by name (e.g. make test-vm-run TEST=TestVM_Journey_DryRun)
+# Single test by name (e.g. make test-vm-run TEST=TestVM_Journey_DryRunIsCompletelySafe)
 test-vm-run: build
 	go test -v -timeout 45m -tags="e2e,vm" -run $(TEST) ./test/e2e/...
 

test/e2e/misc_e2e_test.go

@@ -14,7 +14,7 @@ func TestE2E_FullPreset_DryRun(t *testing.T) {
 		t.Skip("skipping VM test in short mode")
 	}
 
-	vm := testutil.NewTartVM(t)
+	vm := testutil.NewMacHost(t)
 	vmInstallHomebrew(t, vm)
 	bin := vmCopyDevBinary(t, vm)
 

test/e2e/openboot_e2e_test.go

@@ -17,7 +17,7 @@ func TestE2E_DryRunMinimal(t *testing.T) {
 		t.Skip("skipping VM test in short mode")
 	}
 
-	vm := testutil.NewTartVM(t)
+	vm := testutil.NewMacHost(t)
 	vmInstallHomebrew(t, vm)
 	bin := vmCopyDevBinary(t, vm)
 
@@ -35,7 +35,7 @@ func TestE2E_DryRunDeveloper(t *testing.T) {
 		t.Skip("skipping VM test in short mode")
 	}
 
-	vm := testutil.NewTartVM(t)
+	vm := testutil.NewMacHost(t)
 	vmInstallHomebrew(t, vm)
 	bin := vmCopyDevBinary(t, vm)
 
@@ -53,7 +53,7 @@ func TestE2E_SnapshotCapture(t *testing.T) {
 		t.Skip("skipping VM test in short mode")
 	}
 
-	vm := testutil.NewTartVM(t)
+	vm := testutil.NewMacHost(t)
 	vmInstallHomebrew(t, vm)
 	bin := vmCopyDevBinary(t, vm)
 
@@ -71,7 +71,7 @@ func TestE2E_InvalidPreset(t *testing.T) {
 		t.Skip("skipping VM test in short mode")
 	}
 
-	vm := testutil.NewTartVM(t)
+	vm := testutil.NewMacHost(t)
 	vmInstallHomebrew(t, vm)
 	bin := vmCopyDevBinary(t, vm)
 
@@ -91,7 +91,7 @@ func TestE2E_MissingGitConfig(t *testing.T) {
 		t.Skip("skipping VM test in short mode")
 	}
 
-	vm := testutil.NewTartVM(t)
+	vm := testutil.NewMacHost(t)
 	vmInstallHomebrew(t, vm)
 	bin := vmCopyDevBinary(t, vm)
 
@@ -108,7 +108,7 @@ func TestE2E_SnapshotWithOutput(t *testing.T) {
 		t.Skip("skipping VM test in short mode")
 	}
 
-	vm := testutil.NewTartVM(t)
+	vm := testutil.NewMacHost(t)
 	vmInstallHomebrew(t, vm)
 	bin := vmCopyDevBinary(t, vm)
 
@@ -127,7 +127,7 @@ func TestE2E_Diff_ThenClean_DryRun_SameSnapshot(t *testing.T) {
 	}
 
 	// Verify diff and clean produce consistent results from the same snapshot
-	vm := testutil.NewTartVM(t)
+	vm := testutil.NewMacHost(t)
 	vmInstallHomebrew(t, vm)
 	bin := vmCopyDevBinary(t, vm)
 

test/e2e/sync_shell_e2e_test.go

@@ -18,7 +18,7 @@ func TestE2E_Sync_Shell_CaptureShell(t *testing.T) {
 		t.Skip("skipping VM test in short mode")
 	}
 
-	vm := testutil.NewTartVM(t)
+	vm := testutil.NewMacHost(t)
 	installOhMyZsh(t, vm)
 	bin := vmCopyDevBinary(t, vm)
 
@@ -49,7 +49,7 @@ func TestE2E_Sync_Shell_NoPanic(t *testing.T) {
 		t.Skip("skipping VM test in short mode")
 	}
 
-	vm := testutil.NewTartVM(t)
+	vm := testutil.NewMacHost(t)
 	installOhMyZsh(t, vm)
 	bin := vmCopyDevBinary(t, vm)
 

test/e2e/vm_edge_cases_test.go

@@ -27,7 +27,7 @@ func TestVM_Edge_ShellActuallyWorks(t *testing.T) {
 		t.Skip("skipping VM edge case in short mode")
 	}
 
-	vm := testutil.NewTartVM(t)
+	vm := testutil.NewMacHost(t)
 	vmInstallHomebrew(t, vm)
 	bin := vmCopyDevBinary(t, vm)