Merge pull request #242 from askui/chore/default_tools

Adds Default tool Lists for Agents

Author: philipph-askui

pyproject.toml

@@ -183,7 +183,7 @@ unfixable = []
 dummy-variable-rgx = "^(_+|(_+[a-zA-Z0-9_]*[a-zA-Z0-9]+?))$"
 
 [tool.ruff.lint.per-file-ignores]
-"src/askui/agent.py" = ["E501"]
+"src/askui/computer_agent.py" = ["E501"]
 "src/askui/android_agent.py" = ["E501"]
 "src/askui/locators/locators.py" = ["E501"]
 "src/askui/locators/relatable.py" = ["E501", "SLF001"]

src/askui/__init__.py

@@ -7,9 +7,9 @@
 
 os.environ["FASTMCP_EXPERIMENTAL_ENABLE_NEW_OPENAPI_PARSER"] = "true"
 
-from .agent import ComputerAgent, VisionAgent
 from .agent_base import Agent
 from .agent_settings import AgentSettings
+from .computer_agent import ComputerAgent, VisionAgent
 from .locators import Locator
 from .models import (
     Base64ImageSourceParam,

src/askui/agent_base.py

@@ -878,3 +878,7 @@ def __exit__(
         traceback: types.TracebackType | None,
     ) -> None:
         self.close()
+
+    @staticmethod
+    def get_default_tools() -> list[Tool]:
+        return []

src/askui/android_agent.py

@@ -81,23 +81,7 @@ def __init__(
         super().__init__(
             reporter=reporter,
             retry=retry,
-            tools=[
-                AndroidScreenshotTool(),
-                AndroidTapTool(),
-                AndroidTypeTool(),
-                AndroidDragAndDropTool(),
-                AndroidKeyTapEventTool(),
-                AndroidSwipeTool(),
-                AndroidKeyCombinationTool(),
-                AndroidShellTool(),
-                AndroidSelectDeviceBySerialNumberTool(),
-                AndroidSelectDisplayByUniqueIDTool(),
-                AndroidGetConnectedDevicesSerialNumbersTool(),
-                AndroidGetConnectedDisplaysInfosTool(),
-                AndroidGetCurrentConnectedDeviceInfosTool(),
-                ExceptionTool(),
-            ]
-            + (act_tools or []),
+            tools=self.get_default_tools() + (act_tools or []),
             agent_os=self.os,
             settings=settings,
             callbacks=callbacks,
@@ -357,6 +341,25 @@ def set_device_by_serial_number(
         )
         self.os.set_device_by_serial_number(device_sn)
 
+    @staticmethod
+    def get_default_tools() -> list[Tool]:
+        return [
+            AndroidScreenshotTool(),
+            AndroidTapTool(),
+            AndroidTypeTool(),
+            AndroidDragAndDropTool(),
+            AndroidKeyTapEventTool(),
+            AndroidSwipeTool(),
+            AndroidKeyCombinationTool(),
+            AndroidShellTool(),
+            AndroidSelectDeviceBySerialNumberTool(),
+            AndroidSelectDisplayByUniqueIDTool(),
+            AndroidGetConnectedDevicesSerialNumbersTool(),
+            AndroidGetConnectedDisplaysInfosTool(),
+            AndroidGetCurrentConnectedDeviceInfosTool(),
+            ExceptionTool(),
+        ]
+
 
 class AndroidVisionAgent(AndroidAgent):
     def __init__(self, *args, **kwargs) -> None:  # type: ignore

src/askui/agent.py src/askui/computer_agent.pysrc/askui/agent.py renamed to src/askui/computer_agent.py

@@ -92,25 +92,7 @@ def __init__(
         super().__init__(
             reporter=reporter,
             retry=retry,
-            tools=[
-                ExceptionTool(),
-                ComputerGetSystemInfoTool(),
-                ComputerGetMousePositionTool(),
-                ComputerKeyboardPressedTool(),
-                ComputerKeyboardReleaseTool(),
-                ComputerKeyboardTapTool(),
-                ComputerMouseClickTool(),
-                ComputerMouseHoldDownTool(),
-                ComputerMouseReleaseTool(),
-                ComputerMouseScrollTool(),
-                ComputerMoveMouseTool(),
-                ComputerScreenshotTool(),
-                ComputerTypeTool(),
-                ComputerListDisplaysTool(),
-                ComputerRetrieveActiveDisplayTool(),
-                ComputerSetActiveDisplayTool(),
-            ]
-            + (act_tools or []),
+            tools=self.get_default_tools() + (act_tools or []),
             agent_os=self.tools.os,
             settings=settings,
             callbacks=callbacks,
@@ -508,6 +490,27 @@ def cli(
         )
         self.tools.os.run_command(command)
 
+    @staticmethod
+    def get_default_tools() -> list[Tool]:
+        return [
+            ExceptionTool(),
+            ComputerGetSystemInfoTool(),
+            ComputerGetMousePositionTool(),
+            ComputerKeyboardPressedTool(),
+            ComputerKeyboardReleaseTool(),
+            ComputerKeyboardTapTool(),
+            ComputerMouseClickTool(),
+            ComputerMouseHoldDownTool(),
+            ComputerMouseReleaseTool(),
+            ComputerMouseScrollTool(),
+            ComputerMoveMouseTool(),
+            ComputerScreenshotTool(),
+            ComputerTypeTool(),
+            ComputerListDisplaysTool(),
+            ComputerRetrieveActiveDisplayTool(),
+            ComputerSetActiveDisplayTool(),
+        ]
+
 
 class VisionAgent(ComputerAgent):
     def __init__(self, *args, **kwargs) -> None:  # type: ignore

src/askui/web_agent.py

@@ -1,6 +1,6 @@
 from pydantic import ConfigDict, validate_call
 
-from askui.agent import ComputerAgent
+from askui import ComputerAgent
 from askui.agent_settings import AgentSettings
 from askui.models.shared.settings import (
     ActSettings,

tests/e2e/agent/conftest.py

@@ -6,8 +6,7 @@
 import pytest
 from typing_extensions import override
 
-from askui import AgentSettings
-from askui.agent import ComputerAgent
+from askui import AgentSettings, ComputerAgent
 from askui.locators.locators import Locator
 from askui.locators.serializers import AskUiLocatorSerializer
 from askui.model_providers.detection_provider import DetectionProvider

tests/e2e/agent/test_act.py

@@ -1,4 +1,4 @@
-from askui.agent import ComputerAgent
+from askui import ComputerAgent
 
 
 def test_act(

tests/e2e/agent/test_act_caching.py

@@ -4,7 +4,7 @@
 import tempfile
 from pathlib import Path
 
-from askui.agent import ComputerAgent
+from askui import ComputerAgent
 from askui.models.shared.settings import CacheExecutionSettings, CachingSettings
 
 

tests/e2e/agent/test_agent_offset.py

@@ -5,7 +5,7 @@
 
 from PIL import Image as PILImage
 
-from askui.agent import ComputerAgent
+from askui import ComputerAgent
 from askui.locators import Element
 
 if TYPE_CHECKING: