fix: handle restore rollback and spawn id collisions

backend/internal/session/manager.go

@@ -101,6 +101,11 @@ func New(d Deps) *Manager {
 // step eagerly rolls back the steps that already succeeded.
 func (m *Manager) Spawn(ctx context.Context, cfg ports.SpawnConfig) (domain.Session, error) {
 	id := m.newID(cfg)
+	if _, ok, err := m.store.Get(ctx, id); err != nil {
+		return domain.Session{}, fmt.Errorf("spawn %s: check existing: %w", id, err)
+	} else if ok {
+		return domain.Session{}, fmt.Errorf("spawn %s: already exists", id)
+	}
 
 	ws, err := m.workspace.Create(ctx, ports.WorkspaceConfig{
 		ProjectID: cfg.ProjectID,
@@ -315,6 +320,9 @@ func (m *Manager) Restore(ctx context.Context, id domain.SessionID) (domain.Sess
 	}
 	if err := m.lcm.OnSpawnCompleted(ctx, id, outcome); err != nil {
 		m.rollbackRuntime(ctx, handle)
+		if revertErr := m.lcm.OnSpawnInitiated(ctx, rec); revertErr != nil {
+			return domain.Session{}, fmt.Errorf("restore %s: revert after spawn completed failure: %w (original error: %v)", id, revertErr, err)
+		}
 		return domain.Session{}, fmt.Errorf("restore %s: on spawn completed: %w", id, err)
 	}
 	return m.Get(ctx, id)

backend/internal/session/manager_test.go

@@ -121,6 +121,32 @@ func TestSpawn_RuntimeCreateFailure_RollsBack(t *testing.T) {
 	}
 }
 
+func TestSpawn_ExistingSessionIDRejectedBeforeWork(t *testing.T) {
+	h := newHarness("sess-1")
+	ctx := context.Background()
+	if err := h.store.Upsert(ctx, domain.SessionRecord{
+		ID:        "sess-1",
+		ProjectID: testProject,
+		Lifecycle: lc(domain.SessionWorking, domain.ReasonTaskInProgress, domain.PRNone, ""),
+	}); err != nil {
+		t.Fatalf("seed existing row: %v", err)
+	}
+
+	_, err := h.sm.Spawn(ctx, spawnCfg())
+	if err == nil {
+		t.Fatal("spawn: want error for existing session id, got nil")
+	}
+	if len(h.workspace.created) != 0 {
+		t.Error("workspace should not be created when session id already exists")
+	}
+	if len(h.runtime.created) != 0 {
+		t.Error("runtime should not be created when session id already exists")
+	}
+	if h.log.indexOf("OnSpawnInitiated") != -1 || h.log.indexOf("OnSpawnCompleted") != -1 {
+		t.Error("LCM should not be called when session id already exists")
+	}
+}
+
 func TestSpawn_OnSpawnCompletedFailure_RoutesOrphanToErrored(t *testing.T) {
 	h := newHarness("sess-1")
 	ctx := context.Background()
@@ -457,6 +483,11 @@ func TestRestore_OnSpawnCompletedFailure_RollsBackRuntime(t *testing.T) {
 		t.Fatal("restore: want error, got nil")
 	}
 
+	rec, _, _ := h.store.Get(ctx, "sess-1")
+	if got := rec.Lifecycle.Session; got.State != domain.SessionTerminated || got.Reason != domain.ReasonManuallyKilled {
+		t.Fatalf("restore failure should restore terminal lifecycle, got %+v", got)
+	}
+
 	// The runtime created during restore is torn back down so no process is
 	// stranded; the workspace is left intact (it holds the agent's prior work).
 	if len(h.runtime.destroyed) != destroyedBefore+1 {