minitap-mobile-use 2.0.0__py3-none-any.whl → 2.1.0__py3-none-any.whl

minitap/mobile_use/tools/mobile/take_screenshot.py

@@ -1,18 +1,17 @@
-from typing import Optional
-
 from langchain_core.messages import ToolMessage
 from langchain_core.tools import tool
 from langchain_core.tools.base import InjectedToolCallId
 from langgraph.prebuilt import InjectedState
 from langgraph.types import Command
+from minitap.mobile_use.constants import EXECUTOR_MESSAGES_KEY
 from minitap.mobile_use.context import MobileUseContext
 from minitap.mobile_use.controllers.mobile_command_controller import (
     take_screenshot as take_screenshot_controller,
 )
 from minitap.mobile_use.graph.state import State
-from minitap.mobile_use.tools.tool_wrapper import ExecutorMetadata, ToolWrapper
+from minitap.mobile_use.tools.tool_wrapper import ToolWrapper
 from minitap.mobile_use.utils.media import compress_base64_jpeg
-from typing_extensions import Annotated
+from typing import Annotated
 
 
 def get_take_screenshot_tool(ctx: MobileUseContext):
@@ -21,7 +20,6 @@ def get_take_screenshot_tool(ctx: MobileUseContext):
         tool_call_id: Annotated[str, InjectedToolCallId],
         state: Annotated[State, InjectedState],
         agent_thought: str,
-        executor_metadata: Optional[ExecutorMetadata],
     ):
         """
         Take a screenshot of the device.
@@ -42,21 +40,19 @@ def get_take_screenshot_tool(ctx: MobileUseContext):
             if has_failed
             else take_screenshot_wrapper.on_success_fn(),
             additional_kwargs={"error": output} if has_failed else {},
+            status="error" if has_failed else "success",
         )
         updates = {
             "agents_thoughts": [agent_thought],
-            "messages": [tool_message],
+            EXECUTOR_MESSAGES_KEY: [tool_message],
         }
         if compressed_image_base64:
             updates["latest_screenshot_base64"] = compressed_image_base64
         return Command(
-            update=take_screenshot_wrapper.handle_executor_state_fields(
+            update=state.sanitize_update(
                 ctx=ctx,
-                state=state,
-                executor_metadata=executor_metadata,
-                tool_message=tool_message,
-                is_failure=has_failed,
-                updates=updates,
+                update=updates,
+                agent="executor",
             ),
         )
 

minitap/mobile_use/tools/mobile/tap.py

@@ -1,16 +1,15 @@
-from typing import Optional
-
 from langchain_core.messages import ToolMessage
 from langchain_core.tools import tool
 from langchain_core.tools.base import InjectedToolCallId
 from langgraph.prebuilt import InjectedState
 from langgraph.types import Command
+from minitap.mobile_use.constants import EXECUTOR_MESSAGES_KEY
 from minitap.mobile_use.context import MobileUseContext
 from minitap.mobile_use.controllers.mobile_command_controller import SelectorRequest
 from minitap.mobile_use.controllers.mobile_command_controller import tap as tap_controller
 from minitap.mobile_use.graph.state import State
-from minitap.mobile_use.tools.tool_wrapper import ExecutorMetadata, ToolWrapper
-from typing_extensions import Annotated
+from minitap.mobile_use.tools.tool_wrapper import ToolWrapper
+from typing import Annotated
 
 
 def get_tap_tool(ctx: MobileUseContext):
@@ -19,9 +18,8 @@ def get_tap_tool(ctx: MobileUseContext):
         tool_call_id: Annotated[str, InjectedToolCallId],
         state: Annotated[State, InjectedState],
         agent_thought: str,
-        executor_metadata: Optional[ExecutorMetadata],
         selector_request: SelectorRequest,
-        index: Optional[int] = None,
+        index: int | None = None,
     ):
         """
         Taps on a selector.
@@ -35,18 +33,16 @@ def get_tap_tool(ctx: MobileUseContext):
             if has_failed
             else tap_wrapper.on_success_fn(selector_request, index),
             additional_kwargs={"error": output} if has_failed else {},
+            status="error" if has_failed else "success",
         )
         return Command(
-            update=tap_wrapper.handle_executor_state_fields(
+            update=state.sanitize_update(
                 ctx=ctx,
-                state=state,
-                executor_metadata=executor_metadata,
-                tool_message=tool_message,
-                is_failure=has_failed,
-                updates={
+                update={
                     "agents_thoughts": [agent_thought],
-                    "messages": [tool_message],
+                    EXECUTOR_MESSAGES_KEY: [tool_message],
                 },
+                agent="executor",
             ),
         )
 

minitap/mobile_use/tools/mobile/wait_for_animation_to_end.py

@@ -1,18 +1,17 @@
-from typing import Optional
-
 from langchain_core.messages import ToolMessage
 from langchain_core.tools import tool
 from langchain_core.tools.base import InjectedToolCallId
 from langgraph.prebuilt import InjectedState
 from langgraph.types import Command
+from minitap.mobile_use.constants import EXECUTOR_MESSAGES_KEY
 from minitap.mobile_use.context import MobileUseContext
 from minitap.mobile_use.controllers.mobile_command_controller import WaitTimeout
 from minitap.mobile_use.controllers.mobile_command_controller import (
     wait_for_animation_to_end as wait_for_animation_to_end_controller,
 )
 from minitap.mobile_use.graph.state import State
-from minitap.mobile_use.tools.tool_wrapper import ExecutorMetadata, ToolWrapper
-from typing_extensions import Annotated
+from minitap.mobile_use.tools.tool_wrapper import ToolWrapper
+from typing import Annotated
 
 
 def get_wait_for_animation_to_end_tool(ctx: MobileUseContext):
@@ -21,8 +20,7 @@ def get_wait_for_animation_to_end_tool(ctx: MobileUseContext):
         tool_call_id: Annotated[str, InjectedToolCallId],
         state: Annotated[State, InjectedState],
         agent_thought: str,
-        executor_metadata: Optional[ExecutorMetadata],
-        timeout: Optional[WaitTimeout],
+        timeout: WaitTimeout | None,
     ):
         """
         Waits for ongoing animations or videos to finish before continuing.
@@ -43,18 +41,16 @@ def get_wait_for_animation_to_end_tool(ctx: MobileUseContext):
             if has_failed
             else wait_for_animation_to_end_wrapper.on_success_fn(timeout),
             additional_kwargs={"error": output} if has_failed else {},
+            status="error" if has_failed else "success",
         )
         return Command(
-            update=wait_for_animation_to_end_wrapper.handle_executor_state_fields(
+            update=state.sanitize_update(
                 ctx=ctx,
-                state=state,
-                executor_metadata=executor_metadata,
-                tool_message=tool_message,
-                is_failure=has_failed,
-                updates={
+                update={
                     "agents_thoughts": [agent_thought],
-                    "messages": [tool_message],
+                    EXECUTOR_MESSAGES_KEY: [tool_message],
                 },
+                agent="executor",
             ),
         )
 

minitap/mobile_use/tools/tool_wrapper.py

@@ -1,33 +1,11 @@
-from typing import Callable, Optional
+from collections.abc import Callable
 
-from langchain_core.messages import ToolMessage
 from langchain_core.tools import BaseTool
 from pydantic import BaseModel
 from minitap.mobile_use.context import MobileUseContext
-from minitap.mobile_use.graph.state import State
-
-
-class ExecutorMetadata(BaseModel):
-    retrigger: bool
 
 
 class ToolWrapper(BaseModel):
     tool_fn_getter: Callable[[MobileUseContext], BaseTool]
     on_success_fn: Callable[..., str]
     on_failure_fn: Callable[..., str]
-
-    def handle_executor_state_fields(
-        self,
-        ctx: MobileUseContext,
-        state: State,
-        executor_metadata: Optional[ExecutorMetadata],
-        is_failure: bool,
-        tool_message: ToolMessage,
-        updates: dict,
-    ):
-        if executor_metadata is None:
-            return state.sanitize_update(ctx=ctx, update=updates)
-        updates["executor_retrigger"] = executor_metadata.retrigger
-        updates["executor_messages"] = [tool_message]
-        updates["executor_failed"] = is_failure
-        return state.sanitize_update(ctx=ctx, update=updates)

minitap/mobile_use/tools/utils.py

@@ -0,0 +1,86 @@
+from __future__ import annotations
+
+from minitap.mobile_use.context import MobileUseContext
+from minitap.mobile_use.controllers.mobile_command_controller import (
+    CoordinatesSelectorRequest,
+    IdSelectorRequest,
+    SelectorRequestWithCoordinates,
+    tap,
+)
+from minitap.mobile_use.graph.state import State
+from minitap.mobile_use.utils.logger import get_logger
+from minitap.mobile_use.utils.ui_hierarchy import (
+    Point,
+    find_element_by_resource_id,
+    get_bounds_for_element,
+    is_element_focused,
+)
+
+logger = get_logger(__name__)
+
+
+def move_cursor_to_end_if_bounds(
+    ctx: MobileUseContext,
+    state: State,
+    resource_id: str,
+    elt: dict | None = None,
+) -> dict | None:
+    """
+    Best-effort move of the text cursor near the end of the input by tapping the
+    bottom-right area of the focused element (if bounds are available).
+    """
+    if not elt:
+        elt = find_element_by_resource_id(
+            ui_hierarchy=state.latest_ui_hierarchy or [],
+            resource_id=resource_id,
+        )
+    if not elt:
+        return
+
+    bounds = get_bounds_for_element(elt)
+    if not bounds:
+        return elt
+
+    logger.debug("Tapping near the end of the input to move the cursor")
+    bottom_right: Point = bounds.get_relative_point(x_percent=0.99, y_percent=0.99)
+    tap(
+        ctx=ctx,
+        selector_request=SelectorRequestWithCoordinates(
+            coordinates=CoordinatesSelectorRequest(
+                x=bottom_right.x,
+                y=bottom_right.y,
+            ),
+        ),
+    )
+    logger.debug(f"Tapped end of input {resource_id} at ({bottom_right.x}, {bottom_right.y})")
+    return elt
+
+
+def focus_element_if_needed(
+    ctx: MobileUseContext,
+    resource_id: str,
+) -> bool:
+    """
+    Ensures the element identified by `resource_id` is focused.
+    """
+    rich_hierarchy: list[dict] = ctx.hw_bridge_client.get_rich_hierarchy()
+    rich_elt = find_element_by_resource_id(
+        ui_hierarchy=rich_hierarchy,
+        resource_id=resource_id,
+        is_rich_hierarchy=True,
+    )
+    if rich_elt and not is_element_focused(rich_elt):
+        tap(ctx=ctx, selector_request=IdSelectorRequest(id=resource_id))
+        logger.debug(f"Focused (tap) on resource_id={resource_id}")
+        rich_hierarchy = ctx.hw_bridge_client.get_rich_hierarchy()
+        rich_elt = find_element_by_resource_id(
+            ui_hierarchy=rich_hierarchy,
+            resource_id=resource_id,
+            is_rich_hierarchy=True,
+        )
+    if rich_elt and is_element_focused(rich_elt):
+        logger.debug(f"Text input is focused: {resource_id}")
+        return True
+
+    logger.warning(f"Failed to focus resource_id={resource_id}")
+    return False

minitap/mobile_use/utils/cli_helpers.py

@@ -3,10 +3,9 @@ import sys
 from minitap.mobile_use.clients.ios_client import get_ios_devices
 from adbutils import AdbClient
 from rich.console import Console
-from typing import Optional
 
 
-def display_device_status(console: Console, adb_client: Optional[AdbClient] = None):
+def display_device_status(console: Console, adb_client: AdbClient | None = None):
     """Checks for connected devices and displays the status."""
     console.print("\n[bold]📱 Device Status[/bold]")
     devices = None

minitap/mobile_use/utils/cli_selection.py

@@ -1,5 +1,4 @@
 import sys
-from typing import List, Optional
 
 import inquirer
 from rich.console import Console
@@ -8,12 +7,12 @@ from rich.prompt import Prompt
 
 def select_provider_and_model(
     console: Console,
-    available_providers: List[str],
+    available_providers: list[str],
     available_models: dict,
     default_provider: str,
     default_model: str,
-    provider: Optional[str] = None,
-    model: Optional[str] = None,
+    provider: str | None = None,
+    model: str | None = None,
 ) -> tuple[str, str]:
     """
     Interactive selection of LLM provider and model with arrow-key dropdowns when available.
@@ -71,7 +70,7 @@ def select_provider_and_model(
 def _select_from_list(
     console: Console,
     item_type: str,
-    choices: List[str],
+    choices: list[str],
     default: str,
     message: str,
 ) -> str:
@@ -108,7 +107,7 @@ def _select_from_list(
         return _numbered_selection(console, item_type, choices, default)
 
 
-def _numbered_selection(console: Console, item_type: str, choices: List[str], default: str) -> str:
+def _numbered_selection(console: Console, item_type: str, choices: list[str], default: str) -> str:
     """Fallback numbered selection when arrow keys aren't available."""
     choices_text = "\n".join([f"  {i + 1}. {choice}" for i, choice in enumerate(choices)])
     console.print(f"Available {item_type}s:\n{choices_text}")

minitap/mobile_use/utils/decorators.py

@@ -1,6 +1,7 @@
 import asyncio
 from functools import wraps
-from typing import Any, Awaitable, Callable, Optional, TypeVar, cast, overload
+from typing import Any, TypeVar, cast, overload
+from collections.abc import Awaitable, Callable
 
 R = TypeVar("R")
 
@@ -8,9 +9,9 @@ R = TypeVar("R")
 def wrap_with_callbacks_sync(
     fn: Callable[..., R],
     *,
-    before: Optional[Callable[..., None]] = None,
-    on_success: Optional[Callable[[R], None]] = None,
-    on_failure: Optional[Callable[[Exception], None]] = None,
+    before: Callable[..., None] | None = None,
+    on_success: Callable[[R], None] | None = None,
+    on_failure: Callable[[Exception], None] | None = None,
     suppress_exceptions: bool = False,
 ) -> Callable[..., R]:
     @wraps(fn)
@@ -35,9 +36,9 @@ def wrap_with_callbacks_sync(
 def wrap_with_callbacks_async(
     fn: Callable[..., Awaitable[R]],
     *,
-    before: Optional[Callable[..., None]] = None,
-    on_success: Optional[Callable[[R], None]] = None,
-    on_failure: Optional[Callable[[Exception], None]] = None,
+    before: Callable[..., None] | None = None,
+    on_success: Callable[[R], None] | None = None,
+    on_failure: Callable[[Exception], None] | None = None,
     suppress_exceptions: bool = False,
 ) -> Callable[..., Awaitable[R]]:
     @wraps(fn)
@@ -63,9 +64,9 @@ def wrap_with_callbacks_async(
 def wrap_with_callbacks(
     fn: Callable[..., Awaitable[R]],
     *,
-    before: Optional[Callable[[], None]] = ...,
-    on_success: Optional[Callable[[R], None]] = ...,
-    on_failure: Optional[Callable[[Exception], None]] = ...,
+    before: Callable[[], None] | None = ...,
+    on_success: Callable[[R], None] | None = ...,
+    on_failure: Callable[[Exception], None] | None = ...,
     suppress_exceptions: bool = ...,
 ) -> Callable[..., Awaitable[R]]: ...
 
@@ -73,9 +74,9 @@ def wrap_with_callbacks(
 @overload
 def wrap_with_callbacks(
     *,
-    before: Optional[Callable[..., None]] = ...,
-    on_success: Optional[Callable[[Any], None]] = ...,
-    on_failure: Optional[Callable[[Exception], None]] = ...,
+    before: Callable[..., None] | None = ...,
+    on_success: Callable[[Any], None] | None = ...,
+    on_failure: Callable[[Exception], None] | None = ...,
     suppress_exceptions: bool = ...,
 ) -> Callable[[Callable[..., R]], Callable[..., R]]: ...
 
@@ -84,19 +85,19 @@ def wrap_with_callbacks(
 def wrap_with_callbacks(
     fn: Callable[..., R],
     *,
-    before: Optional[Callable[[], None]] = ...,
-    on_success: Optional[Callable[[R], None]] = ...,
-    on_failure: Optional[Callable[[Exception], None]] = ...,
+    before: Callable[[], None] | None = ...,
+    on_success: Callable[[R], None] | None = ...,
+    on_failure: Callable[[Exception], None] | None = ...,
     suppress_exceptions: bool = ...,
 ) -> Callable[..., R]: ...
 
 
 def wrap_with_callbacks(
-    fn: Optional[Callable[..., Any]] = None,
+    fn: Callable[..., Any] | None = None,
     *,
-    before: Optional[Callable[[], None]] = None,
-    on_success: Optional[Callable[[Any], None]] = None,
-    on_failure: Optional[Callable[[Exception], None]] = None,
+    before: Callable[[], None] | None = None,
+    on_success: Callable[[Any], None] | None = None,
+    on_failure: Callable[[Exception], None] | None = None,
     suppress_exceptions: bool = False,
 ) -> Any:
     def decorator(func: Callable[..., Any]) -> Any:

minitap/mobile_use/utils/logger.py

@@ -2,7 +2,6 @@ import logging
 import sys
 from enum import Enum
 from pathlib import Path
-from typing import Optional, Union
 
 from colorama import Fore, Style, init
 
@@ -22,7 +21,7 @@ class MobileUseLogger:
     def __init__(
         self,
         name: str,
-        log_file: Optional[Union[str, Path]] = None,
+        log_file: str | Path | None = None,
         console_level: str = "INFO",
         file_level: str = "DEBUG",
         enable_file_logging: bool = True,
@@ -57,7 +56,7 @@ class MobileUseLogger:
 
         self.logger.addHandler(console_handler)
 
-    def _setup_file_handler(self, log_file: Optional[Union[str, Path]], level: str):
+    def _setup_file_handler(self, log_file: str | Path | None, level: str):
         if log_file is None:
             log_file = Path("logs") / f"{self.name.replace('.', '_')}.log"
 
@@ -118,7 +117,7 @@ _loggers = {}
 
 def get_logger(
     name: str,
-    log_file: Optional[Union[str, Path]] = None,
+    log_file: str | Path | None = None,
     console_level: str = "INFO",
     file_level: str = "DEBUG",
     enable_file_logging: bool = False,

minitap/mobile_use/utils/media.py

@@ -55,7 +55,7 @@ def create_steps_json_from_trace_folder(trace_folder_path: Path):
     steps = []
     for file in trace_folder_path.iterdir():
         if file.suffix == ".json":
-            with open(file, "r", encoding="utf-8", errors="ignore") as f:
+            with open(file, encoding="utf-8", errors="ignore") as f:
                 json_content = f.read()
                 steps.append({"timestamp": int(file.stem), "data": json_content})
 

minitap/mobile_use/utils/recorder.py

@@ -1,9 +1,9 @@
 import base64
 import time
-from pathlib import Path
 
+from colorama import Fore, Style
 from langchain_core.messages import BaseMessage
-from minitap.mobile_use.config import record_events
+
 from minitap.mobile_use.context import MobileUseContext
 from minitap.mobile_use.controllers.mobile_command_controller import take_screenshot
 from minitap.mobile_use.utils.logger import get_logger
@@ -45,11 +45,12 @@ def record_interaction(ctx: MobileUseContext, response: BaseMessage):
     return "Screenshot recorded successfully"
 
 
-def log_agent_thoughts(agents_thoughts: list[str], output_path: Path | None):
-    if len(agents_thoughts) > 0:
-        last_agents_thoughts = agents_thoughts[-1]
-        previous_last_agents_thoughts = agents_thoughts[-2] if len(agents_thoughts) > 1 else None
-        if previous_last_agents_thoughts != last_agents_thoughts:
-            logger.info(f"💭 {last_agents_thoughts}")
-            if output_path:
-                record_events(output_path=output_path, events=agents_thoughts)
+def log_agent_thought(prefix: str, agent_thought: str):
+    if prefix:
+        prefix = prefix[0].upper() + prefix[1:]
+    else:
+        prefix = "New agent thought"
+    logger.info(
+        f"💭 {Fore.LIGHTMAGENTA_EX + Style.BRIGHT}{prefix}{Style.RESET_ALL}: "
+        f"{Fore.LIGHTMAGENTA_EX}{agent_thought}{Style.RESET_ALL}"
+    )

minitap/mobile_use/utils/ui_hierarchy.py

@@ -1,7 +1,47 @@
-from typing import Optional
+from pydantic import BaseModel
 
+from minitap.mobile_use.utils.logger import get_logger
 
-def find_element_by_resource_id(ui_hierarchy: list[dict], resource_id: str) -> Optional[dict]:
+logger = get_logger(__name__)
+
+
+def __find_element_by_ressource_id_in_rich_hierarchy(
+    hierarchy: list[dict], resource_id: str
+) -> dict | None:
+    """
+    Retrieves all the sibling elements for a given resource ID from a nested dictionary.
+
+    Args:
+      hierarchy (dict): The nested dictionary representing the UI hierarchy.
+      resource_id (str): The resource-id to find.
+
+    Returns:
+      list: A list of the sibling elements, or None if the resource_id is not found.
+    """
+    if not hierarchy:
+        return None
+
+    for child in hierarchy:
+        if child.get("attributes", {}).get("resource-id") == resource_id:
+            return child.get("attributes", {})
+
+    for child in hierarchy:
+        result = __find_element_by_ressource_id_in_rich_hierarchy(
+            child.get("children", []), resource_id
+        )
+        if result is not None:
+            return result
+
+    return None
+
+
+def text_input_is_empty(text: str | None, hint_text: str | None) -> bool:
+    return not text or text == hint_text
+
+
+def find_element_by_resource_id(
+    ui_hierarchy: list[dict], resource_id: str, is_rich_hierarchy: bool = False
+) -> dict | None:
     """
     Find a UI element by its resource-id in the UI hierarchy.
 
@@ -13,8 +53,10 @@ def find_element_by_resource_id(ui_hierarchy: list[dict], resource_id: str) -> O
     Returns:
         The complete UI element dictionary if found, None otherwise
     """
+    if is_rich_hierarchy:
+        return __find_element_by_ressource_id_in_rich_hierarchy(ui_hierarchy, resource_id)
 
-    def search_recursive(elements: list[dict]) -> Optional[dict]:
+    def search_recursive(elements: list[dict]) -> dict | None:
         for element in elements:
             if isinstance(element, dict):
                 if element.get("resourceId") == resource_id:
@@ -28,3 +70,56 @@ def find_element_by_resource_id(ui_hierarchy: list[dict], resource_id: str) -> O
         return None
 
     return search_recursive(ui_hierarchy)
+
+
+def is_element_focused(element: dict) -> bool:
+    return element.get("focused", None) == "true"
+
+
+def get_element_text(element: dict, hint_text: bool = False) -> str | None:
+    if hint_text:
+        return element.get("hintText", None)
+    return element.get("text", None)
+
+
+class Point(BaseModel):
+    x: int
+    y: int
+
+
+class ElementBounds(BaseModel):
+    x: int
+    y: int
+    width: int
+    height: int
+
+    def get_center(self) -> Point:
+        return Point(x=self.x + self.width // 2, y=self.y + self.height // 2)
+
+    def get_relative_point(self, x_percent: float, y_percent: float) -> Point:
+        """
+        Returns the coordinates of the point at x_percent of the width and y_percent
+        of the height of the element.
+
+        Ex if x_percent = 0.95 and y_percent = 0.95,
+        the point is at the bottom right of the element:
+        <------>
+        |      |
+        |     x|
+        <------>
+        """
+        return Point(
+            x=int((self.x + self.width) * x_percent),
+            y=int((self.y + self.height) * y_percent),
+        )
+
+
+def get_bounds_for_element(element: dict) -> ElementBounds | None:
+    bounds = element.get("bounds", None)
+    if bounds:
+        try:
+            return ElementBounds(**bounds)
+        except Exception as e:
+            logger.error(f"Failed to validate bounds: {e}")
+            return None
+    return None

{minitap_mobile_use-2.0.0.dist-info → minitap_mobile_use-2.1.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: minitap-mobile-use
-Version: 2.0.0
+Version: 2.1.0
 Summary: AI-powered multi-agent system that automates real Android and iOS devices through low-level control using LangGraph.
 Author: Pierre-Louis Favreau, Jean-Pierre Lo, Nicolas Dehandschoewercker
 License: MIT License
@@ -45,7 +45,7 @@ Requires-Dist: psutil>=5.9.0
 Requires-Dist: ruff==0.5.3 ; extra == 'dev'
 Requires-Dist: pytest==8.4.1 ; extra == 'dev'
 Requires-Dist: pytest-cov==5.0.0 ; extra == 'dev'
-Requires-Python: >=3.10
+Requires-Python: >=3.12
 Project-URL: Homepage, https://minitap.ai/
 Project-URL: Source, https://github.com/minitap-ai/mobile-use
 Provides-Extra: dev
@@ -257,6 +257,16 @@ python ./src/mobile_use/main.py \
 > [!NOTE]  
 > If you haven't configured a specific model, mobile-use will prompt you to choose one from the available options.
 
+## 🔎 Agentic System Overview
+
+<div align="center">
+
+![Graph Visualization](doc/graph.png)
+
+_This diagram is automatically updated from the codebase. This is our current agentic system architecture._
+
+</div>
+
 ## ❤️ Contributing
 
 We love contributions! Whether you're fixing a bug, adding a feature, or improving documentation, your help is welcome. Please read our **[Contributing Guidelines](CONTRIBUTING.md)** to get started.