chore: remove tart-based test-vm targets; L4 now runs on github actions

test-vm, test-vm-run, test-vm-parallel and their VM_A/VM_B variables are
deleted. L4 destructive e2e now runs on macos-14 GitHub Actions runners
(vm-e2e-spike.yml) — each job gets a fresh macOS VM at no extra cost for
public repos. Kept test-vm-inner / test-vm-inner-run for local use and
as the commands the CI workflow invokes.

Updated: Makefile, CLAUDE.md, CONTRIBUTING.md, AGENTS.md, HARNESS.md,
auto-release.yml checklist.

Author: fullstackjam

.github/workflows/auto-release.yml

@@ -176,13 +176,13 @@ jobs:
           **not** auto-tag. Run the destructive e2e suite locally, then
           cut the release manually:
 
-          - [ ] \`make test-vm-parallel\` passes (Apple Silicon + Tart required — see scripts/vm/README.md)
+          - [ ] L4 CI (\`vm-e2e-spike.yml\`) is green on the latest commit on \`main\`
           - [ ] sanity-check the curl|bash smoke and cli-compat results in the most recent test.yml run on main
           - [ ] \`git tag -a ${NEW_TAG} -m "..."\` and \`git push origin ${NEW_TAG}\`
           - [ ] close this issue
 
-          Skipping \`make test-vm-parallel\` is allowed (it is not a hard gate),
-          but \`feat:\` changes carry more risk than \`fix:\` patches.
+          L4 CI is not yet a hard merge gate, but \`feat:\` changes carry
+          more risk than \`fix:\` patches — verify it before tagging.
           EOF
           )
 

.github/workflows/vm-e2e-spike.yml

@@ -6,8 +6,7 @@ on:
   workflow_dispatch:
 
 jobs:
-  # Mirrors VM_A in make test-vm-parallel:
-  # long-running journey tests that install packages and modify system state.
+  # Group A: long-running journey tests that install packages and modify system state.
   group-a:
     runs-on: macos-14
     timeout-minutes: 60
@@ -28,8 +27,7 @@ jobs:
             -run 'TestVM_Journey_FirstTimeUser|TestVM_Journey_DryRunIsCompletelySafe|TestVM_Interactive_InstallScript' \
             ./test/e2e/...
 
-  # Mirrors VM_B in make test-vm-parallel:
-  # dotfiles, macOS defaults, edge cases, sync, and non-destructive e2e.
+  # Group B: dotfiles, macOS defaults, edge cases, sync, and non-destructive e2e.
   group-b:
     runs-on: macos-14
     timeout-minutes: 60

AGENTS.md

@@ -81,8 +81,8 @@ These are loaded automatically when Claude runs in this repo.
 - `git push --force` against `main` or release tags.
 - `git commit --amend` on commits already pushed.
 - `git reset --hard` discarding uncommitted work.
-- Running `make test-vm` (or any other `test-vm-*` target) outside an ephemeral
-  VM — these install real packages.
+- Running `make test-vm-inner` (or `test-vm-inner-run`) outside a throwaway
+  machine — these install real packages onto the current host.
 - Anything that modifies the user's `~/.zshrc`, Homebrew install, or
   macOS `defaults`.
 

CLAUDE.md

@@ -24,8 +24,8 @@ make build-release VERSION=0.25.0    # optimized + UPX
 # Test — full tier table in CONTRIBUTING.md
 make test-unit                       # L1 (~75s) — unit + integration + contract; pre-push hook
 make test-e2e                        # L3 compiled binary
-make test-vm-parallel                # L4 (~14 min) — destructive e2e in 2 parallel Tart VMs; before tagging
-make test-vm                         # L4 serial fallback — single VM, use for debugging
+make test-vm-inner                   # L4 — full destructive e2e suite (runs in CI on macos-14; locally on a spare machine only)
+make test-vm-inner-run TEST=Foo      # L4 — single test
 make test-coverage                   # coverage.out + coverage.html
 
 # Single test

CONTRIBUTING.md

@@ -37,37 +37,24 @@ Tests are split across four tiers. Which one runs where:
 | **L1 Unit + Integration + Contract** | Pure-Go logic with faked `Runner` *plus* real `brew` / `git` / `npm` against temp dirs and real `httptest` servers | `make test-unit` (~75s) | Every push (pre-push hook); CI on push/PR |
 | **L2 Contract schema** | JSON schema validation against [openboot-contract](https://github.com/openbootdotdev/openboot-contract) | (runs in CI only) | CI on push/PR |
 | **L3 E2E binary** | Compiled binary driven by scripts; `-tags=e2e` | `make test-e2e` | CI on release |
-| **L4 VM e2e** | Full destructive suite (`-tags="e2e,vm"`) runs inside an ephemeral Tart VM provisioned by `scripts/vm/run.sh`. Installs real packages, modifies `~/.zshrc`, writes `defaults` — all contained to the throwaway VM. | `make test-vm-parallel` (~14 min, Apple Silicon + Tart required); `make test-vm` is the serial single-VM fallback | **Local only** — convention is to run before tagging a release. No CI gate. |
+| **L4 VM e2e** | Full destructive suite (`-tags="e2e,vm"`). Installs real packages, modifies `~/.zshrc`, writes `defaults`. Each run requires a clean macOS host (Apple Silicon). | `make test-vm-inner` (or single test: `make test-vm-inner-run TEST=Foo`) | **CI** — runs on GitHub Actions `macos-14` runner (every PR via `vm-e2e-spike.yml`). Locally only on a throwaway machine. |
 
 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 cutting a release.
 - **Before pushing:** `make test-unit` (the pre-push hook does this automatically). Requires `brew` / `git` / `npm` on PATH — they are queried read-only against temp dirs, no real installs.
-- **Before tagging a release (convention, not enforced):** `make test-vm-parallel` on an Apple Silicon Mac with Tart installed. See [VM E2E setup](#vm-e2e-setup) below. `auto-release.yml` opens a `release-ready` issue on `feat:` thresholds to nudge you here.
+- **Before tagging a release:** check that the L4 CI job (`vm-e2e-spike.yml`) is green on the latest commit on `main`. `auto-release.yml` opens a `release-ready` issue on `feat:` thresholds to nudge you here.
 
-## VM E2E setup
+## VM E2E
 
-Destructive tests (L4) run inside an ephemeral Tart VM. One-time setup
-on an Apple Silicon Mac:
+L4 tests run on GitHub Actions (`macos-14` runner, Apple Silicon). Each job
+gets a fresh macOS VM — no local setup required.
 
 ```bash
-brew install cirruslabs/cli/tart
-tart pull ghcr.nju.edu.cn/cirruslabs/tahoe-base:latest
-tart clone ghcr.nju.edu.cn/cirruslabs/tahoe-base:latest tahoe-base
+make test-vm-inner                                   # full suite (use on a throwaway machine only)
+make test-vm-inner-run TEST=TestVM_Journey_FirstTimeUser   # single test
 ```
 
-Then:
-
-```bash
-make test-vm-parallel                                # full suite in parallel (~14 min)
-make test-vm                                         # full suite serial fallback (~30 min)
-make test-vm-run TEST=TestVM_Journey_FirstTimeUser   # one test
-OPENBOOT_VM_KEEP=1 make test-vm-parallel             # don't destroy VMs at exit (debug)
-```
-
-See `scripts/vm/README.md` for full environment-variable docs and
-troubleshooting.
-
 ## Git Hooks
 
 `make install-hooks` symlinks two hooks from `scripts/hooks/` into `.git/hooks/`:

Makefile

@@ -1,12 +1,7 @@
 .PHONY: test-unit test-e2e test-coverage test-all \
-       test-vm test-vm-run test-vm-parallel test-vm-inner test-vm-inner-run \
+       test-vm-inner test-vm-inner-run \
        install-hooks uninstall-hooks
 
-# VM A: install/journey tests that touch real system state (longest-running).
-VM_A_TESTS := TestVM_Journey_FirstTimeUser|TestVM_Journey_DryRunIsCompletelySafe|TestVM_Interactive_InstallScript
-# VM B: all other VM tests — dotfiles, macOS, edge cases, smoke, real-install, sync.
-VM_B_TESTS := TestVM_Journey_Dotfiles|TestVM_Journey_MacOS|TestVM_Journey_FullSetupConfiguresEverything|TestVM_Edge_|TestSmoke_|TestE2E_
-
 BINARY_NAME=openboot
 BINARY_PATH=./$(BINARY_NAME)
 VERSION ?= dev
@@ -34,38 +29,17 @@ test-all:
 	$(MAKE) test-coverage
 
 # =============================================================================
-# Tart VM e2e — destructive tests run inside a throwaway Tart VM provisioned
-# by scripts/vm/run.sh. Files tagged `e2e,vm` run via `make test-vm-inner`;
-# files tagged `e2e && !vm` (auth, snapshot_api) run as L3 on the host.
-#
-# Requires Apple Silicon + Tart installed locally. See scripts/vm/README.md
-# for one-time setup. The relevant targets are defined immediately below
-# this header: test-vm, test-vm-run, test-vm-inner, test-vm-inner-run.
+# L4 VM e2e — destructive tests tagged `e2e,vm`. Run directly on any clean
+# macOS host (Apple Silicon). In CI this is a GitHub Actions macos-14 runner
+# (see .github/workflows/vm-e2e-spike.yml). Locally, run on a throwaway
+# machine or a Tart VM — do NOT run on your primary dev machine.
 # =============================================================================
 
-# Developer-facing: provisions a Tart VM and runs the full e2e suite inside.
-test-vm: build
-	scripts/vm/run.sh test-vm-inner
-
-# Developer-facing: runs one named test inside a Tart VM.
-test-vm-run: build
-	scripts/vm/run.sh "test-vm-inner-run TEST=$(TEST)"
-
-# Developer-facing: runs e2e in two parallel VMs — VM A (system tests) and
-# VM B (mock-server tests). Requires ~16 GB RAM and 8 cores free.
-# Exit code is non-zero if either VM fails.
-test-vm-parallel: build
-	@OPENBOOT_VM_TEST='$(VM_A_TESTS)' scripts/vm/run.sh test-vm-inner & PID_A=$$!; \
-	OPENBOOT_VM_TEST='$(VM_B_TESTS)' scripts/vm/run.sh test-vm-inner & PID_B=$$!; \
-	A_EXIT=0; B_EXIT=0; \
-	wait $$PID_A || A_EXIT=$$?; \
-	wait $$PID_B || B_EXIT=$$?; \
-	[ $$A_EXIT -eq 0 ] && [ $$B_EXIT -eq 0 ]
-
-# In-VM: invoked over SSH by run.sh — not called by developers directly.
+# Run the full L4 suite (same command CI uses).
 test-vm-inner:
 	go test -v -timeout 60m -tags="e2e,vm" ./test/e2e/...
 
+# Run a single L4 test by name: make test-vm-inner-run TEST=TestVM_Journey_FirstTimeUser
 test-vm-inner-run:
 	go test -v -timeout 45m -tags="e2e,vm" -run '$(TEST)' ./test/e2e/...
 

docs/HARNESS.md

@@ -48,9 +48,9 @@ Three regulation categories:
 | Behav. | L1 unit + integration + contract (faked runners *and* real brew/git/npm in temp dirs) | pre-push, CI | `make test-unit` |
 | Behav. | L2 contract schema (against openboot-contract repo) | CI | `.github/workflows/test.yml` `contract` job |
 | Behav. | L3 e2e binary | release | `make test-e2e` |
-| Behav. | L4 VM e2e (`vm`) — runs full destructive suite in a local Tart VM | local only (convention is pre-release; no CI gate) | `make test-vm-parallel` (~14 min, driver: `scripts/vm/run.sh`); `make test-vm` is the serial single-VM fallback |
+| Behav. | L4 VM e2e (`vm`) — full destructive suite on a clean macOS host | every PR | `.github/workflows/vm-e2e-spike.yml` (macos-14 runner, two parallel jobs); `make test-vm-inner` for local runs |
 | Behav. | curl\|bash smoke (install.sh + mock server) | every PR | `.github/workflows/test.yml` `curl-bash-smoke` job |
-| Behav. | Auto-release sensor — patch fast lane (`fix:`-only) auto-tags + dispatches `release.yml`; feat threshold opens a `release-ready` issue with a `make test-vm-parallel` checklist instead | push to `main` | `.github/workflows/auto-release.yml` |
+| Behav. | Auto-release sensor — patch fast lane (`fix:`-only) auto-tags + dispatches `release.yml`; feat threshold opens a `release-ready` issue (check L4 CI green, then tag manually) | push to `main` | `.github/workflows/auto-release.yml` |
 | Behav. | Release notes — Conventional Commits since previous tag, grouped by type (Features / Bug Fixes / etc) + Full Changelog link, appended to the install-instructions template | tag push or `workflow_dispatch` | `.github/workflows/release.yml` (`Write release notes` step) |
 | Behav. | Old-CLI compat (previous release × current mock server) | every PR | `.github/workflows/test.yml` `cli-compat` job |
 | Feedfwd. | Agent conventions | every AI turn | `CLAUDE.md`, `AGENTS.md` |
@@ -110,13 +110,12 @@ it survives doc rot.
   to the inline `\r\033[K` renderer when unavailable. A static rule can't
   see runtime terminal capabilities, so this stays a runtime concern. The
   fallback is covered by `TestStickyProgressFallsBackWhenScrollRegionUnsupported`.
-- **No CI gate for VM e2e.** Apple Silicon Tart VMs don't run on
-  GitHub-hosted `macos-latest` runners (no nested virt, wrong arch
-  guarantees), and we declined to set up a self-hosted runner. L4 is
-  local-only. Running `make test-vm-parallel` before tagging is convention,
-  encoded as a `release-ready` issue opened by `auto-release.yml`
-  on `feat:` thresholds — not a hard gate. A human can release without
-  it.
+- **L4 runs on GitHub Actions, not a self-hosted runner.** `macos-14`
+  runners are Apple Silicon VMs — each job gets a fresh clean macOS
+  environment, which is exactly what L4 needs. Tart is no longer required.
+  The L4 workflow (`vm-e2e-spike.yml`) is not yet a hard merge gate (not in
+  `required-checks.txt`); it runs on every PR. Promoting it to a required
+  check is the next step once the workflow has proven stable.
 
 ## How agents should think about this file