fix(executor): periodically sweep stale dispatch workspace directories

ContainerRunner preserves a failed execution's workspace indefinitely for
debugging, with no expiry -- this accumulated 161 stale directories
(~17.6GB) and took the host to 100% disk full on 2026-07-10. Pool.RunWorkspaceCleanup
sweeps claudomator-workspace-* dirs older than 24h every hour, started from
serve.go, mirroring StoryOrchestrator.Run's ticker shape. Never removes a
directory still referenced as a currently-BLOCKED task's sandbox_dir,
regardless of age.

1 files changed, 70 insertions, 0 deletions

diff --git a/internal/executor/executor_test.go b/internal/executor/executor_test.go
index a2228c3..c40f274 100644
--- a/internal/executor/executor_test.go
+++ b/internal/executor/executor_test.go
@@ -2170,3 +2170,73 @@ func TestPool_Shutdown_TimesOut(t *testing.T) {
 	}
 	close(unblock) // cleanup
 }
+
+// makeStaleWorkspaceDir creates a real directory matching the
+// "claudomator-workspace-*" glob cleanupStaleWorkspaces looks for, backdates
+// its mtime by age, and registers it for cleanup at test end regardless of
+// whether the function under test already removed it.
+func makeStaleWorkspaceDir(t *testing.T, age time.Duration) string {
+	t.Helper()
+	dir, err := os.MkdirTemp("", "claudomator-workspace-*")
+	if err != nil {
+		t.Fatalf("MkdirTemp: %v", err)
+	}
+	t.Cleanup(func() { os.RemoveAll(dir) })
+	old := time.Now().Add(-age)
+	if err := os.Chtimes(dir, old, old); err != nil {
+		t.Fatalf("Chtimes: %v", err)
+	}
+	return dir
+}
+
+func TestPool_CleanupStaleWorkspaces_RemovesOldUnreferencedDir(t *testing.T) {
+	store := testStore(t)
+	logger := slog.New(slog.NewTextHandler(os.Stderr, &slog.HandlerOptions{Level: slog.LevelError}))
+	pool := NewPool(2, map[string]Runner{"claude": &mockRunner{}}, store, logger)
+
+	dir := makeStaleWorkspaceDir(t, 48*time.Hour)
+
+	pool.cleanupStaleWorkspaces(24 * time.Hour)
+
+	if _, err := os.Stat(dir); !os.IsNotExist(err) {
+		t.Errorf("expected stale unreferenced workspace dir to be removed, stat err = %v", err)
+	}
+}
+
+func TestPool_CleanupStaleWorkspaces_KeepsRecentDir(t *testing.T) {
+	store := testStore(t)
+	logger := slog.New(slog.NewTextHandler(os.Stderr, &slog.HandlerOptions{Level: slog.LevelError}))
+	pool := NewPool(2, map[string]Runner{"claude": &mockRunner{}}, store, logger)
+
+	dir := makeStaleWorkspaceDir(t, 1*time.Hour)
+
+	pool.cleanupStaleWorkspaces(24 * time.Hour)
+
+	if _, err := os.Stat(dir); err != nil {
+		t.Errorf("expected recent workspace dir to be kept, got stat err = %v", err)
+	}
+}
+
+func TestPool_CleanupStaleWorkspaces_KeepsBlockedTaskSandboxDirRegardlessOfAge(t *testing.T) {
+	store := testStore(t)
+	logger := slog.New(slog.NewTextHandler(os.Stderr, &slog.HandlerOptions{Level: slog.LevelError}))
+	pool := NewPool(2, map[string]Runner{"claude": &mockRunner{}}, store, logger)
+
+	dir := makeStaleWorkspaceDir(t, 30*24*time.Hour) // deliberately far older than maxAge
+
+	tk := makeTask("blocked-with-sandbox")
+	tk.State = task.StateBlocked
+	store.CreateTask(tk)
+	store.CreateExecution(&storage.Execution{
+		ID: "exec-blocked-1", TaskID: tk.ID,
+		StartTime:  time.Now().Add(-30 * 24 * time.Hour),
+		Status:     "BLOCKED",
+		SandboxDir: dir,
+	})
+
+	pool.cleanupStaleWorkspaces(24 * time.Hour)
+
+	if _, err := os.Stat(dir); err != nil {
+		t.Errorf("expected a BLOCKED task's sandbox_dir to be preserved regardless of age, got stat err = %v", err)
+	}
+}