feat(cua): release 0.38.0+artemis.1 with verified shell observation

Upgrade the pinned driver, validate menu and taskbar focus ownership, patch untitled Shell resolution, and remove duplicate scroll scaling. Publish package provenance, verification boundaries and matching marketplace checksum. Go tests, vet, build and byte-level package verification passed.

plugins/cua-driver/main.go

@@ -25,7 +25,7 @@ import (
 //go:embed tools.json
 var toolsJSON []byte
 
-const adapterVersion = "0.36.0+artemis.4"
+const adapterVersion = "0.38.0+artemis.1"
 
 type obj = map[string]any
 type request struct {
@@ -291,11 +291,13 @@ func (a *adapter) observe(args obj) (obj, obj, error) {
 	delete(a.captures, targetKey(args))
 	started := float64(time.Now().UnixMilli()) / 1000
 	before := foregroundWindow()
+	focusBefore := keyboardFocus()
 	r, err := a.peer.tool("get_window_state", args)
 	if err != nil {
 		return nil, nil, err
 	}
 	annotateCaptureFocus(r, num(args["window_id"]), before, foregroundWindow())
+	object(r["structuredContent"])["keyboard_focus"] = obj{"before": focusBefore, "after": keyboardFocus()}
 	src := source(r, args, started)
 	if frame, ok := capturedFrame(r, src); ok {
 		a.captures[targetKey(args)] = frame
@@ -331,6 +333,7 @@ func (a *adapter) handle(name string, args obj) obj {
 		query, _ := windowQuery(args) // Already validated above.
 		r, err = a.peer.tool(name, query)
 		if err == nil {
+			r = supplementShellWindows(r, query, nativeShellWindows())
 			r = windowListing(r, query)
 		}
 	} else {
@@ -346,34 +349,9 @@ func (a *adapter) handle(name string, args obj) obj {
 				delete(a.captures, targetKey(args))
 				return failure("Window moved or resized; get_window_state again", "not_started", "none", "stale_capture")
 			}
-			if name == "scroll" {
-				// v0.36.0 scroll omits screenshot scaling. Measure actual native
-				// capture pixels (not DWM/DPI estimates) without another model call.
-				// Do not expose this internal frame or reuse its tokens.
-				delete(a.captures, targetKey(args))
-				prepared, e := nativeScrollArgs(frame, args, func() (pointerFrame, error) {
-					nativeArgs := observationArgs(args)
-					nativeArgs["max_image_dimension"] = 0
-					nativeArgs["include_accessibility_tree"] = false
-					native, err := a.peer.tool("get_window_state", nativeArgs)
-					if err != nil {
-						return pointerFrame{}, err
-					}
-					src := source(native, args, float64(time.Now().UnixMilli())/1000)
-					f, ok := capturedFrame(native, src)
-					if !ok {
-						return pointerFrame{}, errors.New("native capture unavailable; get_window_state again")
-					}
-					return f, nil
-				})
-				if e != nil {
-					return failure(e.Error(), "not_started", "none", "invalid_coordinates")
-				}
-				args = prepared
-			} else if name == "drag" {
-				// Upstream drag already scales using its current screenshot.
-				args = copyArgs(args)
-				delete(args, "capture_id")
+			if name == "scroll" || name == "drag" {
+				// v0.38.0 maps screenshot pixels itself. Keep x/y unchanged.
+				args = withoutCaptureBinding(args)
 			}
 		}
 		upstream := name
@@ -402,13 +380,40 @@ func (a *adapter) handle(name string, args obj) obj {
 	}
 	if writeTool(name) && rec["execution"] != "not_started" {
 		// Observe once after delivery; NEVER repeat an uncertain write.
-		observation, origin, e := a.observe(observationArgs(args))
+		focus := keyboardFocus()
+		next, reason := postActionTarget(args, nativeWindowInfo(num(args["window_id"])), focus, nativeShellWindows())
+		structured["post_action_focus"] = focus
+		structured["observation_target"] = obj{"requested": obj{"pid": args["pid"], "window_id": args["window_id"]},
+			"observed": obj{"pid": next["pid"], "window_id": next["window_id"]}, "reason": reason}
+		observation, origin, e := a.observe(next)
 		if e == nil && origin != nil {
 			src = origin
 			structured["observation"] = observation["structuredContent"]
 			r["content"] = append(items(r["content"]), items(observation["content"])...)
 		} else {
 			structured["observation_error"] = "Post-action evidence unavailable; inspect before another action."
+			if e != nil {
+				structured["observation_error"] = e.Error()
+			} else if observation != nil {
+				var errors []string
+				if observation["isError"] == true {
+					for _, item := range items(observation["content"]) {
+						block := object(item)
+						if block["type"] == "text" {
+							text := []rune(str(block["text"]))
+							if len(text) > 2000 {
+								text = text[:2000]
+							}
+							errors = append(errors, string(text))
+						}
+					}
+				}
+				// Failed identity/geometry validation must never forward its image
+				// or full tree as task evidence, even inside diagnostics.
+				structured["observation_diagnostics"] = obj{"driver_errors": errors,
+					"native_target": nativeWindowInfo(num(next["window_id"])), "isError": observation["isError"],
+					"image_size": func() []int { w, h := imageSize(observation); return []int{w, h} }()}
+			}
 		}
 	}
 	if name == "verify_state" { // Read-only verification may include a screenshot; it is not a task verdict.