diff --git a/cmd/create/createCluster.go b/cmd/create/createCluster.go
index 22094ae5..2f047c01 100644
--- a/cmd/create/createCluster.go
+++ b/cmd/create/createCluster.go
@@ -84,8 +84,9 @@ func NewCmdCreateCluster() *cobra.Command {
 			if runtime.GOOS == "windows" {
 				log.Debugf("GOOS is %s", runtime.GOOS)
 				fmt.Printf("$env:KUBECONFIG=(%s get kubeconfig %s)\n", os.Args[0], cluster.Name)
+			} else {
+				fmt.Printf("export KUBECONFIG=$(%s get kubeconfig %s)\n", os.Args[0], cluster.Name)
 			}
-			fmt.Printf("export KUBECONFIG=$(%s get kubeconfig %s)\n", os.Args[0], cluster.Name)
 			fmt.Println("kubectl cluster-info")
 		},
 	}
diff --git a/cmd/get/getKubeconfig.go b/cmd/get/getKubeconfig.go
index 6afe4244..42557d50 100644
--- a/cmd/get/getKubeconfig.go
+++ b/cmd/get/getKubeconfig.go
@@ -58,7 +58,7 @@ func NewCmdGetKubeconfig() *cobra.Command {
 	}
 
 	// add flags
-	cmd.Flags().StringP("output", "o", clientcmd.RecommendedHomeFile, "Define output [ - | FILE ]")
+	cmd.Flags().StringP("output", "o", "", fmt.Sprintf("Define output [ - | FILE ] (default from $KUBECONFIG or %s", clientcmd.RecommendedHomeFile))
 	if err := cmd.MarkFlagFilename("output"); err != nil {
 		log.Fatalln("Failed to mark flag --output as filename")
 	}
diff --git a/docs/documentation.md b/docs/documentation.md
index 61d202b5..ffbbd170 100644
--- a/docs/documentation.md
+++ b/docs/documentation.md
@@ -10,3 +10,7 @@
     * all other master nodes will refer to the initializing master node via `--server https://<init-node>:6443`
 * API-Ports
   * by default, we don't expose any API-Port (no host port mapping)
+* kubeconfig
+  * if no output is set explicitly (via the `--output` flag), we use the default loading rules to get the default kubeconfig:
+    * First: kubeconfig specified via the KUBECONFIG environment variable (error out if multiple are specified)
+    * Second: default kubeconfig in home directory (e.g. `$HOME/.kube/config`)
diff --git a/pkg/cluster/kubeconfig.go b/pkg/cluster/kubeconfig.go
index 413dc43d..29c61711 100644
--- a/pkg/cluster/kubeconfig.go
+++ b/pkg/cluster/kubeconfig.go
@@ -54,6 +54,15 @@ func GetAndWriteKubeConfig(runtime runtimes.Runtime, cluster *k3d.Cluster, outpu
 		return err
 	}
 
+	// empty output parameter = write to default
+	if output == "" {
+		defaultKubeConfigLoadingRules := clientcmd.NewDefaultClientConfigLoadingRules()
+		if len(defaultKubeConfigLoadingRules.GetLoadingPrecedence()) > 1 {
+			return fmt.Errorf("Multiple kubeconfigs specified via KUBECONFIG env var: Please reduce to one entry, unset KUBECONFIG or explicitly choose an output")
+		}
+		output = defaultKubeConfigLoadingRules.GetDefaultFilename()
+	}
+
 	// simply write to the output, ignoring existing contents
 	if writeKubeConfigOptions.OverwriteExisting || output == "-" {
 		return WriteKubeConfigToPath(kubeconfig, output)
@@ -176,7 +185,7 @@ func GetKubeconfig(runtime runtimes.Runtime, cluster *k3d.Cluster) (*clientcmdap
 	return kc, nil
 }
 
-// WriteKubeConfigToPath uses GetKubeConfig to grab the kubeconfig from the cluster master node, writes it to a file and outputs the path
+// WriteKubeConfigToPath takes a kubeconfig and writes it to some path, which can be '-' for os.Stdout
 func WriteKubeConfigToPath(kubeconfig *clientcmdapi.Config, path string) error {
 	var output *os.File
 	defer output.Close()
@@ -209,7 +218,7 @@ func WriteKubeConfigToPath(kubeconfig *clientcmdapi.Config, path string) error {
 
 }
 
-// UpdateKubeConfig merges a new kubeconfig into the existing default kubeconfig
+// UpdateKubeConfig merges a new kubeconfig into an existing kubeconfig and returns the result
 func UpdateKubeConfig(newKubeConfig *clientcmdapi.Config, existingKubeConfig *clientcmdapi.Config, outPath string, overwriteConflicting bool, updateCurrentContext bool) error {
 
 	log.Debugf("Merging new KubeConfig:\n%+v\n>>> into existing KubeConfig:\n%+v", newKubeConfig, existingKubeConfig)
@@ -268,10 +277,3 @@ func UpdateKubeConfig(newKubeConfig *clientcmdapi.Config, existingKubeConfig *cl
 
 	return nil
 }
-
-/*func UpdateKubeConfigOrCreate(outPath string) error {
-	_, err := os.Stat(outPath)
-	if os.IsNotExist(err) {
-		Create
-	}
-}*/
diff --git a/thoughts.md b/thoughts.md
index ca93f7e8..161f12b4 100644
--- a/thoughts.md
+++ b/thoughts.md
@@ -39,6 +39,9 @@
       - --no-headers
     - kubeconfig NAME
       - --output
+      - --overwrite
+      - --switch
+      - --update
   - start
     - cluster NAME
       - --all
@@ -72,7 +75,7 @@
     - --env                       -> planned
     - --label                     -> planned
     - --workers                   -> implemented
-    - --auto-restart              -> planned
+    - --auto-restart              -> dropped (docker's `unless-stopped` is set by default)
     - --enable-registry           -> planned (possible consolidation into less registry-related commands?)
     - --registry-name             -> TBD
     - --registry-port             -> TBD
@@ -105,7 +108,7 @@
   - get-kubeconfig                -> `k3d get kubeconfig CLUSTERNAME`
     - --name                      -> dropped, implemented as arg
     - --all                       -> planned
-    - --overwrite                 -> planned
+    - --overwrite                 -> implemented
   - import-images                 -> `k3d load image [--cluster CLUSTERNAME] [--keep] IMAGES`
     - --name                      -> implemented as `--cluster`
     - --no-remove                 -> implemented as `--keep`
@@ -201,9 +204,14 @@ Here's how k3d types should translate to a runtime type:
 
 ## [WIP] Multi-Master Setup
 
-- if `--masters` > 1 deploy a load-balancer in front of them as an extra container
+- to make this possible, we always deploy a load-balancer (nginx) in front of the master nodes as an extra container 
   - consider that in the kubeconfig file and `--tls-san`
-  - make this the default, but provide a `--no-lb` flag
+
+### Variants
+
+- [x] embedded datastore (dqlite)
+  - if `--masters` > 1 deploy a load-balancer in front of them as an extra container
+- [ ] external datastore
 
 ## [DONE] Keep State in Docker Labels
 
@@ -222,7 +230,7 @@ Here's how k3d types should translate to a runtime type:
   - let's you set tools container version
     - `k3d tools --image k3d-tools:v2 import-images`
 - add `k3d create --image-vol NAME` flag to re-use existing image volume
-  - will add `k3d.volumes.imagevolume.external: true` label to nodes
+  - will add `k3d.volumes.imageVolume.external: true` label to nodes
     - should not be deleted with cluster
   - possibly add `k3d create volume` and `k3d create network` to create external volumes/networks?
 
@@ -256,4 +264,3 @@ Here's how k3d types should translate to a runtime type:
 ### Possible Enhancements
 
 - [!] remove/add nodes -> needs to remove line in `/var/lib/rancher/k3s/server/cred/node-passwd` for the deleted node
-- directly use cluster struct in `createClusterCmd` filled by flag values