fix: iOS spec driving, InputText replace semantics, native DoubleTap (#58)

* fix(hierarchy): bounds-containment fallback for scoped and path queries

Compose on iOS surfaces a testTag node as an empty leaf sibling of the
content it labels instead of as an ancestor, so descendant search under
the tagged node finds nothing and every path or scoped query returns
null. When structural search yields no match, fall back to nodes whose
bounds lie inside the scope node's bounds.

* feat(sidecar): derive iOS clickable and editable from element type

The XCTest hierarchy mapping dropped the element type, leaving no
clickable or editable flags on iOS, so the fuzzer's tap and typing
verbs never found a candidate inside the app. Map the raw
accessibility tree directly and derive clickable, editable,
scrollable, and class from the XCUIElementType raw value.

* feat(proto): add EraseText RPC for InputText replace semantics

* feat(driver): add EraseText to the device driver surface

* fix(runner): erase existing field text before InputText

InputText appended on native platforms, so repeated draws grew fields
without bound. The folio fuzz run wedged on the add-account screen:
each draw concatenated another name until the 40-character validation
error became permanent. Replace semantics also makes retried typing
idempotent. The web driver already replaced via select-all; native now
matches.

* feat(sidecar): EraseText backend support on android and ios

* fix(folio): saturation-gate account creation in the spec

The 2-3 step add-account loop outcompeted the 5-step transaction chain
at every weighted re-draw, so runs filled with account creation and
rarely exercised the balance properties. Stop offering add-account once
three accounts exist; the renormalized weights then favor the
transaction flow at every step of its chain.

* fix(folio): author spec weights to match testing intent

Revert the account saturation gate: it starved newAccountBalanceIsZero
once it tripped, and a magic account count is app-state tuning, not
intent. Instead weight the generators by what the properties need:
the transaction chain leads, account creation stays exercised, and
doubleTaps gets explicit weight everywhere because double-submission
idempotency is what the spec is testing for.

* fix(folio): lower doubleTaps weight to 5

* fix(sidecar): surface visible text on iOS static elements

Static text and button strings live in the accessibility label on
iOS, so the text attribute came through empty and every balance
extractor parsed to zero, silently disarming both folio properties.
Non-editable elements now fall back title, value, then label;
editable fields keep value-only so an empty field's caption does not
read as content.

* feat(driver): native DoubleTap RPC for a tight inter-tap gap

Composing two Tap round trips from the Go client spread the taps by
hundreds of milliseconds on iOS, wide enough for the app to navigate
between them, so double-submission races could never reproduce. The
sidecar now lands both taps back-to-back next to the device transport.

* feat(sidecar): pipeline iOS double-tap requests

Queue the second tap at the XCTest runner while the first executes.
The runner serializes handlers, so this is the tightest gap the
transport allows (~350ms per tap round trip); recorded here with
measurements for the iOS double-tap limitation.

internal/driver/chrome/driver.go

@@ -193,6 +193,23 @@ func (d *Driver) InputText(_ context.Context, text string) error {
 	)
 }
 
+// EraseText clears the focused field. InputText above already replaces via
+// select-all, so the character count is not needed to bound the deletion.
+func (d *Driver) EraseText(_ context.Context, _ int) error {
+	return chromedp.Run(d.tabCtx,
+		chromedp.ActionFunc(func(ctx context.Context) error {
+			if err := chromedp.Evaluate(`
+				(function() {
+					const el = document.activeElement;
+					if (el && typeof el.select === 'function') el.select();
+				})()`, nil).Do(ctx); err != nil {
+				return err
+			}
+			return input.InsertText("").Do(ctx)
+		}),
+	)
+}
+
 func (d *Driver) Swipe(_ context.Context, fromX, fromY, toX, toY int, duration time.Duration) error {
 	millis := max(duration.Milliseconds(), 50)
 	script := fmt.Sprintf(`

internal/driver/driver.go

@@ -20,6 +20,10 @@ type DeviceDriver interface {
 	DoubleTap(ctx context.Context, x, y int) error
 	DoubleTapSelector(ctx context.Context, selector string) error
 	InputText(ctx context.Context, text string) error
+	// EraseText deletes characterCount characters from the focused field.
+	// The runner calls it before InputText so the verb replaces existing
+	// content instead of appending to it.
+	EraseText(ctx context.Context, characterCount int) error
 	Swipe(ctx context.Context, fromX, fromY, toX, toY int, duration time.Duration) error
 	PressKey(ctx context.Context, key string) error
 	LongPress(ctx context.Context, x, y int) error

internal/driver/mock/mock.go

@@ -19,6 +19,7 @@ const (
 	ActionDoubleTap         ActionKind = "double_tap"
 	ActionDoubleTapSelector ActionKind = "double_tap_selector"
 	ActionInputText         ActionKind = "input_text"
+	ActionEraseText         ActionKind = "erase_text"
 	ActionSwipe             ActionKind = "swipe"
 	ActionPressKey          ActionKind = "press_key"
 	ActionLongPress         ActionKind = "long_press"
@@ -32,19 +33,20 @@ const (
 )
 
 type Action struct {
-	Kind         ActionKind
-	BundleID     string
-	ClearState   bool
-	X, Y         int
-	FromX, FromY int
-	ToX, ToY     int
-	Duration     time.Duration
-	Selector     string
-	Text         string
-	Key          string
-	LogLevel     string
-	LogSince     time.Time
-	Idle         time.Duration
+	Kind           ActionKind
+	BundleID       string
+	ClearState     bool
+	X, Y           int
+	FromX, FromY   int
+	ToX, ToY       int
+	Duration       time.Duration
+	Selector       string
+	Text           string
+	CharacterCount int
+	Key            string
+	LogLevel       string
+	LogSince       time.Time
+	Idle           time.Duration
 }
 
 // Driver is an in-memory Driver implementation for unit tests.
@@ -210,6 +212,14 @@ func (d *Driver) InputText(ctx context.Context, text string) error {
 	return nil
 }
 
+func (d *Driver) EraseText(ctx context.Context, characterCount int) error {
+	if err := d.failure(ActionEraseText); err != nil {
+		return err
+	}
+	d.record(Action{Kind: ActionEraseText, CharacterCount: characterCount})
+	return nil
+}
+
 func (d *Driver) Swipe(ctx context.Context, fromX, fromY, toX, toY int, duration time.Duration) error {
 	if err := d.failure(ActionSwipe); err != nil {
 		return err

internal/driver/sidecar/client.go

@@ -104,14 +104,18 @@ func (c *Client) TapSelector(ctx context.Context, selector string) error {
 	return err
 }
 
-// doubleTapGap is the inter-tap delay for DoubleTap: short enough to land both
-// events inside a sub-100 ms race window, long enough for the sidecar to
-// serialize two MotionEvent streams. The sidecar exposes no native double-tap
-// RPC, so the gesture is two Taps with this gap.
+// doubleTapGap is the inter-tap delay for the selector fallback: short enough
+// to land both events inside a sub-100 ms race window, long enough for the
+// sidecar to serialize two MotionEvent streams.
 const doubleTapGap = 50 * time.Millisecond
 
+// DoubleTap dispatches the native RPC so the backend can land both taps as
+// close together as the platform allows. Composing two Tap round trips from
+// here spreads them by hundreds of milliseconds on iOS, wide enough for
+// navigation to interleave between the taps.
 func (c *Client) DoubleTap(ctx context.Context, x, y int) error {
-	return doubleTap(ctx, func() error { return c.Tap(ctx, x, y) })
+	_, err := c.stub.DoubleTap(ctx, &driverpb.Point{X: int32(x), Y: int32(y)})
+	return err
 }
 
 func (c *Client) DoubleTapSelector(ctx context.Context, selector string) error {
@@ -137,6 +141,11 @@ func (c *Client) InputText(ctx context.Context, text string) error {
 	return err
 }
 
+func (c *Client) EraseText(ctx context.Context, characterCount int) error {
+	_, err := c.stub.EraseText(ctx, &driverpb.EraseTextRequest{CharacterCount: int32(characterCount)})
+	return err
+}
+
 func (c *Client) Swipe(ctx context.Context, fromX, fromY, toX, toY int, duration time.Duration) error {
 	_, err := c.stub.Swipe(ctx, &driverpb.SwipeRequest{
 		From:           &driverpb.Point{X: int32(fromX), Y: int32(fromY)},

internal/hierarchy/hierarchy.go

@@ -75,6 +75,7 @@ type Element struct {
 type Node struct {
 	Element
 	Children []*Node `json:"-"`
+	tree     *Tree
 }
 
 // Tree is a flat collection of every node in a hierarchy dump, in pre-order.
@@ -180,7 +181,7 @@ func Parse(text string) (*Tree, error) {
 }
 
 func walkNode(node *treeNodeJSON, tree *Tree) *Node {
-	n := &Node{Element: *elementFromNode(node)}
+	n := &Node{Element: *elementFromNode(node), tree: tree}
 	tree.Elements = append(tree.Elements, &n.Element)
 	for i := range node.Children {
 		n.Children = append(n.Children, walkNode(&node.Children[i], tree))
@@ -333,69 +334,50 @@ func (t *Tree) FindAllBySelectorPath(path []Selector) []*Node {
 	return t.Root.FindAllBySelectorPath(path)
 }
 
-// Find returns the first Node in this node's subtree (descendants only) matching
-// the string selector. Path queries within the selector are not supported here.
+// Find returns the first Node scoped to this node (descendants, with spatial
+// fallback) matching the string selector. Path queries within the selector are
+// not supported here.
 func (n *Node) Find(selector string) *Node {
-	kind, value, ok := parseSelector(selector)
-	if !ok {
-		return nil
-	}
-	for _, child := range n.Children {
-		if nodes := searchSubtree(child, kind, value); len(nodes) > 0 {
-			return nodes[0]
-		}
-	}
-	return nil
+	return firstNode(n.FindAll(selector))
 }
 
-// FindAll returns all Nodes in this node's subtree (descendants only) matching
-// the string selector.
+// FindAll returns all Nodes scoped to this node (descendants, with spatial
+// fallback) matching the string selector.
 func (n *Node) FindAll(selector string) []*Node {
 	kind, value, ok := parseSelector(selector)
 	if !ok {
 		return nil
 	}
-	var result []*Node
-	for _, child := range n.Children {
-		result = append(result, searchSubtree(child, kind, value)...)
-	}
-	return result
+	return n.scopedNodes(func(element *Element) bool {
+		return match(element, kind, value)
+	})
 }
 
-// FindBySelector returns the first Node in this node's subtree matching sel (AND semantics).
+// FindBySelector returns the first Node scoped to this node matching sel (AND semantics).
 func (n *Node) FindBySelector(sel Selector) *Node {
-	for _, child := range n.Children {
-		if nodes := searchSubtreeBySelector(child, sel); len(nodes) > 0 {
-			return nodes[0]
-		}
-	}
-	return nil
+	return firstNode(n.FindAllBySelector(sel))
 }
 
-// FindAllBySelector returns all Nodes in this node's subtree matching sel (AND semantics).
+// FindAllBySelector returns all Nodes scoped to this node matching sel (AND semantics).
 func (n *Node) FindAllBySelector(sel Selector) []*Node {
-	var result []*Node
-	for _, child := range n.Children {
-		result = append(result, searchSubtreeBySelector(child, sel)...)
-	}
-	return result
+	return n.scopedNodes(func(element *Element) bool {
+		return matchSelector(element, sel)
+	})
 }
 
 // FindBySelectorPath walks a chain of selectors. The first selector is matched
-// against descendants of the receiver; each subsequent selector is matched
-// against descendants of the previous match. Returns the deepest match or nil.
+// in the receiver's scope; each subsequent selector is matched in the scope of
+// the previous match. Returns the deepest match or nil.
 func (n *Node) FindBySelectorPath(path []Selector) *Node {
 	if len(path) == 0 {
 		return nil
 	}
-	for _, child := range n.Children {
-		for _, candidate := range searchSubtreeBySelector(child, path[0]) {
-			if len(path) == 1 {
-				return candidate
-			}
-			if deeper := candidate.FindBySelectorPath(path[1:]); deeper != nil {
-				return deeper
-			}
+	for _, candidate := range n.FindAllBySelector(path[0]) {
+		if len(path) == 1 {
+			return candidate
+		}
+		if deeper := candidate.FindBySelectorPath(path[1:]); deeper != nil {
+			return deeper
 		}
 	}
 	return nil
@@ -407,19 +389,86 @@ func (n *Node) FindAllBySelectorPath(path []Selector) []*Node {
 	if len(path) == 0 {
 		return nil
 	}
+	var result []*Node
+	for _, candidate := range n.FindAllBySelector(path[0]) {
+		if len(path) == 1 {
+			result = append(result, candidate)
+			continue
+		}
+		result = append(result, candidate.FindAllBySelectorPath(path[1:])...)
+	}
+	return result
+}
+
+func firstNode(nodes []*Node) *Node {
+	if len(nodes) == 0 {
+		return nil
+	}
+	return nodes[0]
+}
+
+// scopedNodes returns this node's descendants matching accept, in pre-order.
+// When no descendant matches, nodes spatially contained in this node's bounds
+// are matched instead. Compose on iOS emits a testTag node as an empty leaf
+// sibling of the content it labels rather than as an ancestor, so descendant
+// search under the tagged node finds nothing; bounds containment recovers the
+// intended scope.
+func (n *Node) scopedNodes(accept func(*Element) bool) []*Node {
 	var result []*Node
 	for _, child := range n.Children {
-		for _, candidate := range searchSubtreeBySelector(child, path[0]) {
-			if len(path) == 1 {
-				result = append(result, candidate)
-				continue
-			}
-			result = append(result, candidate.FindAllBySelectorPath(path[1:])...)
+		collectMatches(child, accept, &result)
+	}
+	if len(result) > 0 {
+		return result
+	}
+	for _, candidate := range n.spatialScope() {
+		if accept(&candidate.Element) {
+			result = append(result, candidate)
 		}
 	}
 	return result
 }
 
+func collectMatches(node *Node, accept func(*Element) bool, result *[]*Node) {
+	if accept(&node.Element) {
+		*result = append(*result, node)
+	}
+	for _, child := range node.Children {
+		collectMatches(child, accept, result)
+	}
+}
+
+// spatialScope returns every node in the tree, in pre-order, whose positive
+// bounds lie fully inside this node's bounds, excluding the node itself.
+func (n *Node) spatialScope() []*Node {
+	if n.tree == nil || n.tree.Root == nil {
+		return nil
+	}
+	if n.Bounds.Width() <= 0 || n.Bounds.Height() <= 0 {
+		return nil
+	}
+	var result []*Node
+	var walk func(*Node)
+	walk = func(candidate *Node) {
+		if candidate != n &&
+			candidate.Bounds.Width() > 0 && candidate.Bounds.Height() > 0 &&
+			containsBounds(n.Bounds, candidate.Bounds) {
+			result = append(result, candidate)
+		}
+		for _, child := range candidate.Children {
+			walk(child)
+		}
+	}
+	walk(n.tree.Root)
+	return result
+}
+
+// containsBounds reports whether outer fully contains inner (inclusive).
+func containsBounds(outer, inner Bounds) bool {
+	return inner.Left >= outer.Left && inner.Top >= outer.Top &&
+		inner.Right <= outer.Right && inner.Bottom <= outer.Bottom
+}
+
 func findPathNode(root *Node, segments []string) *Node {
 	if root == nil || len(segments) == 0 {
 		return nil
@@ -440,18 +489,12 @@ func findPathNode(root *Node, segments []string) *Node {
 }
 
 func findPathDescendantsNode(root *Node, segments []string) *Node {
-	kind, value, ok := parseSelector(segments[0])
-	if !ok {
-		return nil
-	}
-	for _, child := range root.Children {
-		for _, node := range searchSubtree(child, kind, value) {
-			if len(segments) == 1 {
-				return node
-			}
-			if result := findPathDescendantsNode(node, segments[1:]); result != nil {
-				return result
-			}
+	for _, node := range root.FindAll(segments[0]) {
+		if len(segments) == 1 {
+			return node
+		}
+		if result := findPathDescendantsNode(node, segments[1:]); result != nil {
+			return result
 		}
 	}
 	return nil
@@ -477,19 +520,13 @@ func findPathAllNodes(root *Node, segments []string) []*Node {
 }
 
 func findPathAllDescendantsNodes(root *Node, segments []string) []*Node {
-	kind, value, ok := parseSelector(segments[0])
-	if !ok {
-		return nil
-	}
 	var result []*Node
-	for _, child := range root.Children {
-		for _, node := range searchSubtree(child, kind, value) {
-			if len(segments) == 1 {
-				result = append(result, node)
-				continue
-			}
-			result = append(result, findPathAllDescendantsNodes(node, segments[1:])...)
+	for _, node := range root.FindAll(segments[0]) {
+		if len(segments) == 1 {
+			result = append(result, node)
+			continue
 		}
+		result = append(result, findPathAllDescendantsNodes(node, segments[1:])...)
 	}
 	return result
 }

internal/hierarchy/hierarchy_test.go

@@ -730,3 +730,145 @@ func TestExplicitPackageAttributeWins(t *testing.T) {
 		t.Errorf("package = %q, want app.folio (explicit attr should win)", got)
 	}
 }
+
+// iosFlatDump mirrors the Compose-on-iOS accessibility shape: the testTag node
+// surfaces as an empty leaf SIBLING of the container holding the content it
+// labels, with equal bounds, instead of as an ancestor.
+const iosFlatDump = `{
+  "attributes": {"bounds": "[0,0][402,874]"},
+  "children": [
+    {
+      "attributes": {"accessibilityText": "Folio", "bounds": "[0,0][402,874]"},
+      "children": [
+        {
+          "attributes": {"resource-id": "LoginScreen", "bounds": "[0,62][402,840]"},
+          "children": []
+        },
+        {
+          "attributes": {"bounds": "[0,62][402,840]"},
+          "children": [
+            {
+              "attributes": {"accessibilityText": "EMAIL", "bounds": "[20,106][60,120]"},
+              "children": []
+            },
+            {
+              "attributes": {"resource-id": "LoginEmail", "accessibilityText": "Email", "bounds": "[34,125][368,173]"},
+              "children": [
+                {"attributes": {"bounds": "[34,125][368,173]"}, "children": []}
+              ]
+            },
+            {
+              "attributes": {"resource-id": "LoginPassword", "accessibilityText": "Password", "bounds": "[34,205][368,253]"},
+              "children": []
+            },
+            {
+              "attributes": {"text": "Sign in", "bounds": "[20,297][382,345]"},
+              "children": []
+            },
+            {
+              "attributes": {"resource-id": "LoginSubmit", "accessibilityText": "Sign in", "bounds": "[20,297][382,345]"},
+              "children": [
+                {"attributes": {"text": "Sign in", "resource-id": "SubmitLabel", "bounds": "[168,312][233,330]"}, "children": []}
+              ]
+            }
+          ]
+        }
+      ]
+    },
+    {
+      "attributes": {"bounds": "[0,0][402,54]"},
+      "children": [
+        {
+          "attributes": {"resource-id": "StatusClock", "accessibilityText": "3:24 PM", "bounds": "[55,22][92,42]"},
+          "children": []
+        }
+      ]
+    }
+  ]
+}`
+
+func TestIOSFlatSelectorPathFallsBackToBounds(t *testing.T) {
+	tree, err := Parse(iosFlatDump)
+	if err != nil {
+		t.Fatalf("Parse: %v", err)
+	}
+	path := []Selector{
+		{Filters: []AttrFilter{{Attr: "testTag", Value: "LoginScreen"}}},
+		{Filters: []AttrFilter{{Attr: "testTag", Value: "LoginEmail"}}},
+	}
+	node := tree.FindBySelectorPath(path)
+	if node == nil {
+		t.Fatal("expected LoginEmail via bounds containment under leaf LoginScreen")
+	}
+	if node.ResourceID != "LoginEmail" {
+		t.Fatalf("got %q, want LoginEmail", node.ResourceID)
+	}
+}
+
+func TestIOSFlatStringPathFallsBackToBounds(t *testing.T) {
+	tree, _ := Parse(iosFlatDump)
+	element := tree.Find("id:LoginScreen > id:LoginSubmit")
+	if element == nil {
+		t.Fatal("expected LoginSubmit via bounds containment under leaf LoginScreen")
+	}
+	if element.ResourceID != "LoginSubmit" {
+		t.Fatalf("got %q, want LoginSubmit", element.ResourceID)
+	}
+}
+
+func TestIOSFlatScopedNodeFindFallsBackToBounds(t *testing.T) {
+	tree, _ := Parse(iosFlatDump)
+	screen := tree.FindNode("id:LoginScreen")
+	if screen == nil {
+		t.Fatal("expected LoginScreen node")
+	}
+	node := screen.Find("id:LoginPassword")
+	if node == nil {
+		t.Fatal("expected LoginPassword via bounds containment")
+	}
+	if node.ResourceID != "LoginPassword" {
+		t.Fatalf("got %q, want LoginPassword", node.ResourceID)
+	}
+}
+
+func TestIOSFlatFindAllBySelectorPathReturnsEachField(t *testing.T) {
+	tree, _ := Parse(iosFlatDump)
+	path := []Selector{
+		{Filters: []AttrFilter{{Attr: "testTag", Value: "LoginScreen"}}},
+		{Filters: []AttrFilter{{Attr: "label", Value: "word"}}},
+	}
+	// label "word" substring-matches "Password" only.
+	nodes := tree.FindAllBySelectorPath(path)
+	if len(nodes) != 1 {
+		t.Fatalf("got %d nodes, want 1", len(nodes))
+	}
+	if nodes[0].ResourceID != "LoginPassword" {
+		t.Fatalf("got %q, want LoginPassword", nodes[0].ResourceID)
+	}
+}
+
+func TestIOSFlatSpatialScopeExcludesOutsideBounds(t *testing.T) {
+	tree, _ := Parse(iosFlatDump)
+	// StatusClock sits in the status bar above LoginScreen's bounds; the
+	// spatial fallback must not leak it into the screen's scope.
+	if tree.Find("id:LoginScreen > id:StatusClock") != nil {
+		t.Fatal("StatusClock is outside LoginScreen bounds, expected nil")
+	}
+}
+
+func TestIOSFlatStructuralChildStillPreferred(t *testing.T) {
+	tree, _ := Parse(iosFlatDump)
+	submit := tree.FindNode("id:LoginSubmit")
+	if submit == nil {
+		t.Fatal("expected LoginSubmit node")
+	}
+	// "Sign in" exists as the structural child of LoginSubmit and as an
+	// equal-bounds sibling decoration; the structural child must win.
+	node := submit.Find("text:Sign in")
+	if node == nil {
+		t.Fatal("expected structural child match")
+	}
+	if node.ResourceID != "SubmitLabel" {
+		t.Fatalf("expected the structural child SubmitLabel, got id=%q", node.ResourceID)
+	}
+}

internal/runner/runner.go

@@ -497,6 +497,15 @@ func applyAction(ctx context.Context, drv driver.DeviceDriver, action verifier.A
 				return err
 			}
 		}
+		// InputText replaces the field's content: erase what the target
+		// holds before typing. Appending instead lets repeated draws grow
+		// the field without bound (e.g. into a max-length validation error
+		// the fuzzer can never escape) and makes retried typing land twice.
+		if count := existingTextLength(action, tree); count > 0 {
+			if err := drv.EraseText(ctx, count); err != nil {
+				return err
+			}
+		}
 		return drv.InputText(ctx, action.Text)
 	case verifier.ActionKindSwipe:
 		duration := time.Duration(action.DurationMillis) * time.Millisecond
@@ -547,6 +556,20 @@ func collectLogs(ctx context.Context, drv driver.DeviceDriver, since time.Time)
 	return result
 }
 
+// existingTextLength returns the character count of the InputText target's
+// current text, so the runner can erase it before typing. Zero when the
+// target cannot be resolved or holds no text.
+func existingTextLength(action verifier.Action, tree *hierarchy.Tree) int {
+	if action.On == "" || tree == nil {
+		return 0
+	}
+	element := tree.Find(action.On)
+	if element == nil {
+		return 0
+	}
+	return len([]rune(element.Text))
+}
+
 func resolveCoordinates(action verifier.Action, tree *hierarchy.Tree) (int, int, bool) {
 	// When On is empty, X/Y are authoritative (web V8 path emits coordinates
 	// directly from getBoundingClientRect; the runtime nullifies unresolved

internal/runner/runner_test.go

@@ -370,6 +370,52 @@ func TestRunner_LogsWaitForIdleDriverErrors(t *testing.T) {
 	}
 }
 
+func TestApplyAction_InputTextErasesExistingTextBeforeTyping(t *testing.T) {
+	tree, err := hierarchy.Parse(`{"attributes":{"resource-id":"root","bounds":"[0,0,1080,2340]"},"children":[
+		{"attributes":{"resource-id":"username","text":"stale-value","bounds":"[10,10,500,100]"},"children":[]}
+	]}`)
+	if err != nil {
+		t.Fatalf("Parse: %v", err)
+	}
+	driverMock := mockdriver.New()
+	action := verifier.Action{Kind: verifier.ActionKindInputText, On: "id:username", Text: "alice"}
+
+	if err := applyAction(context.Background(), driverMock, action, tree); err != nil {
+		t.Fatalf("applyAction: %v", err)
+	}
+	actions := driverMock.Actions()
+	if len(actions) != 3 {
+		t.Fatalf("want tap, erase, input; got %v", actions)
+	}
+	if actions[0].Kind != mockdriver.ActionTap {
+		t.Errorf("first action = %v, want tap", actions[0].Kind)
+	}
+	if actions[1].Kind != mockdriver.ActionEraseText || actions[1].CharacterCount != len("stale-value") {
+		t.Errorf("second action = %+v, want erase_text of %d characters", actions[1], len("stale-value"))
+	}
+	if actions[2].Kind != mockdriver.ActionInputText || actions[2].Text != "alice" {
+		t.Errorf("third action = %+v, want input_text alice", actions[2])
+	}
+}
+
+func TestApplyAction_InputTextSkipsEraseWhenTargetEmpty(t *testing.T) {
+	tree, err := hierarchy.Parse(`{"attributes":{"resource-id":"root","bounds":"[0,0,1080,2340]"},"children":[
+		{"attributes":{"resource-id":"username","bounds":"[10,10,500,100]"},"children":[]}
+	]}`)
+	if err != nil {
+		t.Fatalf("Parse: %v", err)
+	}
+	driverMock := mockdriver.New()
+	action := verifier.Action{Kind: verifier.ActionKindInputText, On: "id:username", Text: "alice"}
+
+	if err := applyAction(context.Background(), driverMock, action, tree); err != nil {
+		t.Fatalf("applyAction: %v", err)
+	}
+	if containsAction(driverMock.Actions(), mockdriver.ActionEraseText, "") {
+		t.Errorf("empty field must not be erased: %v", driverMock.Actions())
+	}
+}
+
 func TestApplyAction_InputTextSurfacesFocusTapError(t *testing.T) {
 	t.Run("selector focus tap fails", func(t *testing.T) {
 		driverMock := mockdriver.New()