Initial commit from agent

capabilities/screen_vision.py

@@ -0,0 +1,48 @@
+"""
+Screen vision — captures screenshots and identifies active/open windows.
+Uses mss for fast capture and win32gui for window enumeration.
+"""
+import mss
+import base64
+import io
+from PIL import Image
+
+try:
+    import win32gui
+    WIN32_AVAILABLE = True
+except ImportError:
+    WIN32_AVAILABLE = False
+    print("[JARVIS] pywin32 not available — window detection disabled")
+
+def get_active_window_title() -> str:
+    if WIN32_AVAILABLE:
+        return win32gui.GetWindowText(win32gui.GetForegroundWindow())
+    return "Unknown"
+
+def list_open_windows() -> list[str]:
+    if not WIN32_AVAILABLE:
+        return []
+    windows = []
+    def _enum(hwnd, _):
+        if win32gui.IsWindowVisible(hwnd):
+            title = win32gui.GetWindowText(hwnd)
+            if title:
+                windows.append(title)
+    win32gui.EnumWindows(_enum, None)
+    return windows
+
+def capture_screen() -> str:
+    """Returns base64-encoded PNG of the primary monitor."""
+    with mss.mss() as sct:
+        monitor = sct.monitors[1]
+        img = sct.grab(monitor)
+        pil_img = Image.frombytes("RGB", img.size, img.bgra, "raw", "BGRX")
+        buffer = io.BytesIO()
+        pil_img.save(buffer, format="PNG")
+        return base64.b64encode(buffer.getvalue()).decode()
+
+def capture_screen_file(path: str = "screenshot.png") -> str:
+    """Saves screenshot to file and returns the path."""
+    with mss.mss() as sct:
+        sct.shot(mon=1, output=path)
+    return path