feat: add configurable screenshot resolution limit

tests/unit/admin_console/test_model_settings.py

@@ -27,6 +27,7 @@ def isolated_settings(monkeypatch, tmp_path):
         "ARTEMIS_VIDEO_ENABLED",
         "ARTEMIS_OUTPUTTER_ENABLED",
         "ARTEMIS_STEP_SUMMARIZER_ENABLED",
+        "ARTEMIS_SCREENSHOT_LIMIT_ENABLED",
     ]
     for key in keys:
         monkeypatch.setattr(settings, key, getattr(settings, key))
@@ -93,21 +94,38 @@ def test_rejects_invalid_base_url(url):
 def test_capabilities_persist_and_apply_to_new_tasks(isolated_settings):
     result = service.save_capability_settings(
         service.CapabilitySettings(
-            video_enabled=False, outputter_enabled=False, step_summarizer_enabled=False
+            video_enabled=False,
+            outputter_enabled=False,
+            step_summarizer_enabled=False,
+            screenshot_limit_enabled=False,
         )
     )
     assert result.video_enabled is False
     saved = dotenv_values(isolated_settings)
     assert saved["ARTEMIS_VIDEO_ENABLED"] == "false"
+    assert saved["ARTEMIS_SCREENSHOT_LIMIT_ENABLED"] == "false"
     config = load_agent_config()
     assert config.video_analyzer.enabled is False
     assert config.pro.video_analyzer.enabled is False
     assert config.outputter.enabled is False
     assert config.flash.step_summarizer.enabled is False
+    assert config.screenshot.enabled is False
     service.save_capability_settings(service.CapabilitySettings(video_enabled=True))
     assert load_agent_config().video_analyzer.enabled is True
 
 
+def test_screenshot_json_can_override_environment(monkeypatch, tmp_path):
+    config_path = tmp_path / "artemis.jsonc"
+    config_path.write_text(
+        '{"agent": {"screenshot": {"enabled": true, "max_long_edge": 1024}}}',
+        encoding="utf-8",
+    )
+    monkeypatch.setenv("ARTEMIS_SCREENSHOT_LIMIT_ENABLED", "false")
+    screenshot = load_agent_config(config_path).screenshot
+    assert screenshot.enabled is True
+    assert screenshot.max_long_edge == 1024
+
+
 def test_failed_write_keeps_running_settings(monkeypatch, isolated_settings):
     previous = settings.OPENAI_MODEL
     monkeypatch.setattr(

tests/unit/test_drivers.py

@@ -15,14 +15,18 @@
 """Unit tests for Unified Base Device Driver and implementations."""
 
 import base64
+from io import BytesIO
 from types import SimpleNamespace
 from unittest.mock import AsyncMock, MagicMock
 
 import pytest
+from PIL import Image
 
 from artemis.drivers.base import KeyCode, ScreenData, SwipeDirection
 from artemis.drivers.mock.mock_driver import MockDeviceDriver
 from artemis.drivers.android.adb_driver import AndroidAdbDriver
+from artemis.config.agent import ScreenshotConfig
+from artemis.utils.screenshot_limit import limit_screenshot_long_edge
 
 
 _ONE_PIXEL_PNG = base64.b64encode(
@@ -32,6 +36,13 @@ _ONE_PIXEL_PNG = base64.b64encode(
 ).decode("ascii")
 
 
+def test_screenshot_limit_keeps_small_and_invalid_images_unchanged():
+    small_image = base64.b64decode(_ONE_PIXEL_PNG)
+    assert limit_screenshot_long_edge(small_image, 2048) is small_image
+    invalid_image = b"not an image"
+    assert limit_screenshot_long_edge(invalid_image, 2048) is invalid_image
+
+
 @pytest.mark.asyncio
 async def test_mock_driver_actions():
     """Verify MockDeviceDriver records and performs all core actions."""
@@ -141,6 +152,36 @@ async def test_android_driver_preserves_elements_from_combined_screen_data():
     mock_ui_client.get_hierarchy.assert_not_called()
 
 
+@pytest.mark.asyncio
+@pytest.mark.parametrize("enabled", [True, False])
+async def test_android_screenshot_limit_preserves_native_coordinates(enabled):
+    image = Image.new("RGB", (1200, 2600), "white")
+    buffer = BytesIO()
+    image.save(buffer, format="JPEG")
+    original = buffer.getvalue()
+    elements = [{"text": "目标", "bounds": "[100,200][300,400]"}]
+    ui_client = MagicMock()
+    ui_client.get_screen_data.return_value = SimpleNamespace(
+        base64=base64.b64encode(original).decode("ascii"),
+        hierarchy_xml="<hierarchy />",
+        elements=elements,
+        width=1200,
+        height=2600,
+    )
+    driver = AndroidAdbDriver("emulator-5554", MagicMock(), ui_adb_client=ui_client)
+    driver._screenshot_config = ScreenshotConfig(enabled=enabled, max_long_edge=2048)
+
+    screen = await driver.get_screen_data(skip_settling=True)
+
+    with Image.open(BytesIO(screen.screenshot_bytes)) as result:
+        assert result.size == ((945, 2048) if enabled else (1200, 2600))
+    assert screen.width == 1200 and screen.height == 2600
+    assert screen.ui_elements == elements
+    assert screen.screenshot_base64 == base64.b64encode(screen.screenshot_bytes).decode("ascii")
+    if not enabled:
+        assert screen.screenshot_bytes == original
+
+
 @pytest.mark.asyncio
 async def test_android_driver_parses_xml_hierarchy_fallback():
     """XML-only clients remain supported without discarding their hierarchy."""