oagi-core 0.9.0__tar.gz → 0.9.1__tar.gz

{oagi_core-0.9.0 → oagi_core-0.9.1}/Makefile

@@ -54,4 +54,5 @@ version:
 	@echo "Updating version to $(VERSION) in all files..."
 	@sed -i '' 's/^version = ".*"/version = "$(VERSION)"/' pyproject.toml
 	@sed -i '' 's/^version = ".*"/version = "$(VERSION)"/' metapackage/pyproject.toml
-	@sed -i '' 's/oagi-core\[desktop,server\]==.*/oagi-core[desktop,server]==$(VERSION)",/' metapackage/pyproject.toml
+	@sed -i '' 's/oagi-core\[desktop,server\]==.*/oagi-core[desktop,server]==$(VERSION)",/' metapackage/pyproject.toml
+	make build-all

{oagi_core-0.9.0 → oagi_core-0.9.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: oagi-core
-Version: 0.9.0
+Version: 0.9.1
 Summary: Official API of OpenAGI Foundation
 Project-URL: Homepage, https://github.com/agiopen-org/oagi
 Author-email: OpenAGI Foundation <contact@agiopen.org>
@@ -28,6 +28,7 @@ License: MIT License
 Requires-Python: >=3.10
 Requires-Dist: httpx>=0.28.0
 Requires-Dist: pydantic>=2.0.0
+Requires-Dist: rich>=13.0.0
 Provides-Extra: desktop
 Requires-Dist: pillow>=11.3.0; extra == 'desktop'
 Requires-Dist: pyautogui>=0.9.54; extra == 'desktop'
@@ -75,22 +76,6 @@ export OAGI_API_KEY="your-api-key"
 export OAGI_BASE_URL="https://api.oagi.com"  # or your server URL
 ```
 
-### Single-Step Analysis
-
-Analyze a screenshot and get recommended actions:
-
-```python
-from oagi import single_step
-
-step = single_step(
-    task_description="Click the submit button",
-    screenshot="screenshot.png"  # or bytes, or Image object
-)
-
-print(f"Actions: {step.actions}")
-print(f"Complete: {step.is_complete}")
-```
-
 ### Automated Task Execution
 
 Run tasks automatically with screenshot capture and action execution:
@@ -142,9 +127,6 @@ config = ImageConfig(
     height=700
 )
 compressed = image.transform(config)
-
-# Use with single_step
-step = single_step("Click button", screenshot=compressed)
 ```
 
 ### Async Support
@@ -153,16 +135,9 @@ Use async client for non-blocking operations and better concurrency:
 
 ```python
 import asyncio
-from oagi import async_single_step, AsyncShortTask
+from oagi import AsyncShortTask
 
 async def main():
-    # Single-step async analysis
-    step = await async_single_step(
-        "Find the search bar",
-        screenshot="screenshot.png"
-    )
-    print(f"Found {len(step.actions)} actions")
-    
     # Async task automation
     task = AsyncShortTask()
     async with task:
@@ -176,7 +151,6 @@ asyncio.run(main())
 
 See the [`examples/`](examples/) directory for more usage patterns:
 - `google_weather.py` - Basic task execution with `ShortTask`
-- `single_step.py` - Basic single-step inference
 - `screenshot_with_config.py` - Image compression and optimization
 - `execute_task_auto.py` - Automated task execution
 - `socketio_server_basic.py` - Socket.IO server example

{oagi_core-0.9.0 → oagi_core-0.9.1}/README.md

@@ -35,22 +35,6 @@ export OAGI_API_KEY="your-api-key"
 export OAGI_BASE_URL="https://api.oagi.com"  # or your server URL
 ```
 
-### Single-Step Analysis
-
-Analyze a screenshot and get recommended actions:
-
-```python
-from oagi import single_step
-
-step = single_step(
-    task_description="Click the submit button",
-    screenshot="screenshot.png"  # or bytes, or Image object
-)
-
-print(f"Actions: {step.actions}")
-print(f"Complete: {step.is_complete}")
-```
-
 ### Automated Task Execution
 
 Run tasks automatically with screenshot capture and action execution:
@@ -102,9 +86,6 @@ config = ImageConfig(
     height=700
 )
 compressed = image.transform(config)
-
-# Use with single_step
-step = single_step("Click button", screenshot=compressed)
 ```
 
 ### Async Support
@@ -113,16 +94,9 @@ Use async client for non-blocking operations and better concurrency:
 
 ```python
 import asyncio
-from oagi import async_single_step, AsyncShortTask
+from oagi import AsyncShortTask
 
 async def main():
-    # Single-step async analysis
-    step = await async_single_step(
-        "Find the search bar",
-        screenshot="screenshot.png"
-    )
-    print(f"Found {len(step.actions)} actions")
-    
     # Async task automation
     task = AsyncShortTask()
     async with task:
@@ -136,7 +110,6 @@ asyncio.run(main())
 
 See the [`examples/`](examples/) directory for more usage patterns:
 - `google_weather.py` - Basic task execution with `ShortTask`
-- `single_step.py` - Basic single-step inference
 - `screenshot_with_config.py` - Image compression and optimization
 - `execute_task_auto.py` - Automated task execution
 - `socketio_server_basic.py` - Socket.IO server example

{oagi_core-0.9.0 → oagi_core-0.9.1}/examples/execute_task_manual.py

@@ -7,20 +7,20 @@
 # -----------------------------------------------------------------------------
 
 from oagi import (
+    Actor,
+    AsyncActor,
     AsyncPyautoguiActionHandler,
     AsyncScreenshotMaker,
-    AsyncTask,
     PyautoguiActionHandler,
     ScreenshotMaker,
-    Task,
 )
 
 
 def execute_task_manual(task_desc, max_steps=5):
     # set OAGI_API_KEY and OAGI_BASE_URL
-    # or ShortTask(api_key="your_api_key", base_url="your_base_url")
-    task = Task()
-    task.init_task(task_desc, max_steps=max_steps)
+    # or Actor(api_key="your_api_key", base_url="your_base_url")
+    actor = Actor()
+    actor.init_task(task_desc, max_steps=max_steps)
     executor = (
         PyautoguiActionHandler()
     )  # executor = lambda actions: print(actions) for debugging
@@ -36,8 +36,8 @@ def execute_task_manual(task_desc, max_steps=5):
         image = image_provider()
 
         # For additional instructions
-        # step = task.step(image, instruction="some instruction")
-        step = task.step(image)
+        # step = actor.step(image, instruction="some instruction")
+        step = actor.step(image)
 
         # do something with step, maybe print to debug
         print(f"Step {i}: {step.reason=}")
@@ -60,9 +60,9 @@ def execute_task_manual(task_desc, max_steps=5):
 
 async def async_execute_task_manual(task_desc, max_steps=5):
     # set OAGI_API_KEY and OAGI_BASE_URL
-    # or ShortTask(api_key="your_api_key", base_url="your_base_url")
-    async with AsyncTask() as task:
-        await task.init_task(task_desc, max_steps=max_steps)
+    # or AsyncActor(api_key="your_api_key", base_url="your_base_url")
+    async with AsyncActor() as actor:
+        await actor.init_task(task_desc, max_steps=max_steps)
         executor = AsyncPyautoguiActionHandler()
 
         # by default, screenshot will be resized to 1260 * 700 and jpeg with quality 85
@@ -76,8 +76,8 @@ async def async_execute_task_manual(task_desc, max_steps=5):
             image = await image_provider()
 
             # For additional instructions
-            # step = task.step(image, instruction="some instruction")
-            step = await task.step(image)
+            # step = actor.step(image, instruction="some instruction")
+            step = await actor.step(image)
 
             # do something with step, maybe print to debug
             print(f"Step {i}: {step.reason=}")

{oagi_core-0.9.0 → oagi_core-0.9.1}/examples/screenshot_with_config.py

@@ -6,7 +6,7 @@
 #  Licensed under the MIT License.
 # -----------------------------------------------------------------------------
 
-from oagi import ImageConfig, PILImage, ScreenshotMaker, single_step
+from oagi import ImageConfig, PILImage, ScreenshotMaker
 
 
 def example_full_png_screenshot():
@@ -53,33 +53,9 @@ def example_load_and_compress(file_name):
     return compressed_image
 
 
-def example_with_single_step(file_name):
-    """Example 3: Use compressed image with single_step."""
-    print("\nExample 3: Use with single_step")
-
-    # Load and compress image
-    image = PILImage.from_file(file_name)
-    print(f"Original image dimensions: {image.image.size}")
-
-    config = ImageConfig(format="JPEG", quality=85, width=1260, height=700)
-    compressed = image.transform(config)
-    print(f"Compressed image dimensions: {compressed.image.size}")
-
-    # Use with single_step
-    step = single_step(
-        task_description="Click the submit button",
-        screenshot=compressed,
-        api_key="your-api-key-here",
-        base_url="http://127.0.0.1:8000",
-    )
-
-    print(f"Task complete: {step.is_complete}")
-    return step
-
-
 def example_default_config():
-    """Example 4: Default configuration (1260x700 JPEG with 85 quality)."""
-    print("\nExample 4: Default configuration")
+    """Example 3: Default configuration (1260x700 JPEG with 85 quality)."""
+    print("\nExample 3: Default configuration")
 
     default_maker = ScreenshotMaker()  # Uses default ImageConfig
     default_screenshot = default_maker()

{oagi_core-0.9.0 → oagi_core-0.9.1}/metapackage/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "oagi"
-version = "0.9.0"
+version = "0.9.1"
 description = "Official API of OpenAGI Foundation (metapackage with all features)"
 readme = "README.md"
 license = "MIT"
@@ -16,7 +16,7 @@ authors = [
 requires-python = ">= 3.10"
 
 dependencies = [
-    "oagi-core[desktop,server]==0.9.0",
+    "oagi-core[desktop,server]==0.9.1",
 ]
 
 [project.urls]

{oagi_core-0.9.0 → oagi_core-0.9.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "oagi-core"
-version = "0.9.0"
+version = "0.9.1"
 description = "Official API of OpenAGI Foundation"
 readme = "README.md"
 license = { file = "LICENSE" }
@@ -17,6 +17,7 @@ requires-python = ">= 3.10"
 dependencies = [
     "httpx>=0.28.0",
     "pydantic>=2.0.0",
+    "rich>=13.0.0",
 ]
 
 [project.urls]

{oagi_core-0.9.0 → oagi_core-0.9.1}/src/oagi/__init__.py

@@ -7,7 +7,6 @@
 # -----------------------------------------------------------------------------
 import importlib
 
-from oagi.async_single_step import async_single_step
 from oagi.client import AsyncClient, SyncClient
 from oagi.exceptions import (
     APIError,
@@ -21,8 +20,7 @@ from oagi.exceptions import (
     ServerError,
     ValidationError,
 )
-from oagi.single_step import single_step
-from oagi.task import AsyncShortTask, AsyncTask, ShortTask, Task
+from oagi.task import Actor, AsyncActor, AsyncShortTask, AsyncTask, ShortTask, Task
 from oagi.types import (
     AsyncActionHandler,
     AsyncImageProvider,
@@ -33,12 +31,12 @@ from oagi.types.models import ErrorDetail, ErrorResponse, LLMResponse
 # Lazy imports for pyautogui-dependent modules
 # These will only be imported when actually accessed
 _LAZY_IMPORTS = {
-    "AsyncPyautoguiActionHandler": "oagi.async_pyautogui_action_handler",
-    "AsyncScreenshotMaker": "oagi.async_screenshot_maker",
-    "PILImage": "oagi.pil_image",
-    "PyautoguiActionHandler": "oagi.pyautogui_action_handler",
-    "PyautoguiConfig": "oagi.pyautogui_action_handler",
-    "ScreenshotMaker": "oagi.screenshot_maker",
+    "AsyncPyautoguiActionHandler": "oagi.handler.async_pyautogui_action_handler",
+    "AsyncScreenshotMaker": "oagi.handler.async_screenshot_maker",
+    "PILImage": "oagi.handler.pil_image",
+    "PyautoguiActionHandler": "oagi.handler.pyautogui_action_handler",
+    "PyautoguiConfig": "oagi.handler.pyautogui_action_handler",
+    "ScreenshotMaker": "oagi.handler.screenshot_maker",
     # Agent modules (to avoid circular imports)
     "TaskerAgent": "oagi.agent.tasker",
     # Server modules (optional - requires server dependencies)
@@ -59,18 +57,17 @@ def __getattr__(name: str):
 
 __all__ = [
     # Core sync classes
-    "Task",
-    "ShortTask",
+    "Actor",
+    "AsyncActor",
+    "Task",  # Deprecated: Use Actor instead
+    "ShortTask",  # Deprecated
     "SyncClient",
     # Core async classes
-    "AsyncTask",
-    "AsyncShortTask",
+    "AsyncTask",  # Deprecated: Use AsyncActor instead
+    "AsyncShortTask",  # Deprecated
     "AsyncClient",
     # Agent classes
     "TaskerAgent",
-    # Functions
-    "single_step",
-    "async_single_step",
     # Async protocols
     "AsyncActionHandler",
     "AsyncImageProvider",

{oagi_core-0.9.0 → oagi_core-0.9.1}/src/oagi/agent/default.py

@@ -8,10 +8,11 @@
 
 import logging
 
-from .. import AsyncTask
+from .. import AsyncActor
 from ..types import (
     AsyncActionHandler,
     AsyncImageProvider,
+    AsyncStepObserver,
 )
 
 logger = logging.getLogger(__name__)
@@ -27,12 +28,14 @@ class AsyncDefaultAgent:
         model: str = "lux-v1",
         max_steps: int = 30,
         temperature: float | None = None,
+        step_observer: AsyncStepObserver | None = None,
     ):
         self.api_key = api_key
         self.base_url = base_url
         self.model = model
         self.max_steps = max_steps
         self.temperature = temperature
+        self.step_observer = step_observer
 
     async def execute(
         self,
@@ -40,11 +43,11 @@ class AsyncDefaultAgent:
         action_handler: AsyncActionHandler,
         image_provider: AsyncImageProvider,
     ) -> bool:
-        async with AsyncTask(
+        async with AsyncActor(
             api_key=self.api_key, base_url=self.base_url, model=self.model
-        ) as self.task:
+        ) as self.actor:
             logger.info(f"Starting async task execution: {instruction}")
-            await self.task.init_task(instruction, max_steps=self.max_steps)
+            await self.actor.init_task(instruction, max_steps=self.max_steps)
 
             for i in range(self.max_steps):
                 logger.debug(f"Executing step {i + 1}/{self.max_steps}")
@@ -53,15 +56,28 @@ class AsyncDefaultAgent:
                 image = await image_provider()
 
                 # Get next step from OAGI
-                step = await self.task.step(image, temperature=self.temperature)
+                step = await self.actor.step(image, temperature=self.temperature)
 
                 # Log reasoning
                 if step.reason:
-                    logger.debug(f"Step {i + 1} reasoning: {step.reason}")
+                    logger.info(f"Step {i + 1}: {step.reason}")
+
+                # Notify observer if present
+                if self.step_observer and step.actions:
+                    await self.step_observer.on_step(i + 1, step.reason, step.actions)
 
                 # Execute actions if any
                 if step.actions:
-                    logger.debug(f"Executing {len(step.actions)} actions")
+                    logger.info(f"Actions ({len(step.actions)}):")
+                    for action in step.actions:
+                        count_suffix = (
+                            f" x{action.count}"
+                            if action.count and action.count > 1
+                            else ""
+                        )
+                        logger.info(
+                            f"  [{action.type.value}] {action.argument}{count_suffix}"
+                        )
                     await action_handler(step.actions)
 
                 # Check if task is complete

{oagi_core-0.9.0 → oagi_core-0.9.1}/src/oagi/agent/factories.py

@@ -6,6 +6,7 @@
 #  Licensed under the MIT License.
 # -----------------------------------------------------------------------------
 from oagi.agent.tasker import TaskerAgent
+from oagi.types import AsyncStepObserver
 
 from .default import AsyncDefaultAgent
 from .protocol import AsyncAgent
@@ -19,6 +20,7 @@ def create_default_agent(
     model: str = "lux-v1",
     max_steps: int = 20,
     temperature: float = 0.1,
+    step_observer: AsyncStepObserver | None = None,
 ) -> AsyncAgent:
     return AsyncDefaultAgent(
         api_key=api_key,
@@ -26,6 +28,7 @@ def create_default_agent(
         model=model,
         max_steps=max_steps,
         temperature=temperature,
+        step_observer=step_observer,
     )
 
 
@@ -37,6 +40,7 @@ def create_planner_agent(
     max_steps: int = 30,
     temperature: float = 0.0,
     reflection_interval: int = 20,
+    step_observer: AsyncStepObserver | None = None,
 ) -> AsyncAgent:
     tasker = TaskerAgent(
         api_key=api_key,
@@ -45,6 +49,7 @@ def create_planner_agent(
         max_steps=max_steps,
         temperature=temperature,
         reflection_interval=reflection_interval,
+        step_observer=step_observer,
     )
     # tasker.set_task()
     return tasker

{oagi_core-0.9.0 → oagi_core-0.9.1}/src/oagi/agent/tasker/taskee_agent.py

@@ -10,8 +10,8 @@ import logging
 from datetime import datetime
 from typing import Any
 
-from oagi import AsyncTask
-from oagi.types import AsyncActionHandler, AsyncImageProvider
+from oagi import AsyncActor
+from oagi.types import AsyncActionHandler, AsyncImageProvider, AsyncStepObserver
 
 from ..protocol import AsyncAgent
 from .memory import PlannerMemory
@@ -42,6 +42,7 @@ class TaskeeAgent(AsyncAgent):
         planner: Planner | None = None,
         external_memory: PlannerMemory | None = None,
         todo_index: int | None = None,
+        step_observer: AsyncStepObserver | None = None,
     ):
         """Initialize the taskee agent.
 
@@ -55,6 +56,7 @@ class TaskeeAgent(AsyncAgent):
             planner: Planner for planning and reflection
             external_memory: External memory from parent agent
             todo_index: Index of the todo being executed
+            step_observer: Optional observer for step tracking
         """
         self.api_key = api_key
         self.base_url = base_url
@@ -65,9 +67,10 @@ class TaskeeAgent(AsyncAgent):
         self.planner = planner or Planner()
         self.external_memory = external_memory
         self.todo_index = todo_index
+        self.step_observer = step_observer
 
         # Internal state
-        self.task: AsyncTask | None = None
+        self.actor: AsyncActor | None = None
         self.current_todo: str = ""
         self.current_instruction: str = ""
         self.actions: list[Action] = []
@@ -135,10 +138,10 @@ class TaskeeAgent(AsyncAgent):
             )
             return False
         finally:
-            # Clean up task
-            if self.task:
-                await self.task.close()
-                self.task = None
+            # Clean up actor
+            if self.actor:
+                await self.actor.close()
+                self.actor = None
 
     async def _initial_plan(self, image_provider: AsyncImageProvider) -> None:
         """Generate initial plan for the todo.
@@ -199,17 +202,17 @@ class TaskeeAgent(AsyncAgent):
         logger.info(f"Executing subtask with max {max_steps} steps")
 
         # Use async with for automatic resource management
-        async with AsyncTask(
+        async with AsyncActor(
             api_key=self.api_key,
             base_url=self.base_url,
             model=self.model,
             temperature=self.temperature,
-        ) as task:
+        ) as actor:
             # Store reference for potential cleanup in execute's finally block
-            self.task = task
+            self.actor = actor
 
-            # Initialize task with current instruction
-            await task.init_task(self.current_instruction)
+            # Initialize actor with current instruction
+            await actor.init_task(self.current_instruction)
 
             steps_taken = 0
             for step_num in range(max_steps):
@@ -218,7 +221,7 @@ class TaskeeAgent(AsyncAgent):
 
                 # Get next step from OAGI
                 try:
-                    step = await task.step(screenshot, instruction=None)
+                    step = await actor.step(screenshot, instruction=None)
                 except Exception as e:
                     logger.error(f"Error getting step from OAGI: {e}")
                     self._record_action(
@@ -228,8 +231,24 @@ class TaskeeAgent(AsyncAgent):
                     )
                     break
 
+                # Log reasoning
+                if step.reason:
+                    logger.info(f"Step {self.total_actions + 1}: {step.reason}")
+
                 # Record OAGI actions
                 if step.actions:
+                    # Log actions with details
+                    logger.info(f"Actions ({len(step.actions)}):")
+                    for action in step.actions:
+                        count_suffix = (
+                            f" x{action.count}"
+                            if action.count and action.count > 1
+                            else ""
+                        )
+                        logger.info(
+                            f"  [{action.type.value}] {action.argument}{count_suffix}"
+                        )
+
                     for action in step.actions:
                         self._record_action(
                             action_type=action.type.lower(),
@@ -237,6 +256,12 @@ class TaskeeAgent(AsyncAgent):
                             reasoning=step.reason,
                         )
 
+                    # Notify observer if present
+                    if self.step_observer:
+                        await self.step_observer.on_step(
+                            self.total_actions + 1, step.reason, step.actions
+                        )
+
                     # Execute actions
                     await action_handler(step.actions)
                     self.total_actions += len(step.actions)
@@ -255,9 +280,9 @@ class TaskeeAgent(AsyncAgent):
                     logger.info("Reflection interval reached")
                     break
 
-            # Task will be automatically closed by async with context manager
+            # Actor will be automatically closed by async with context manager
             # Clear reference after context manager closes it
-            self.task = None
+            self.actor = None
             return steps_taken
 
     async def _reflect_and_decide(self, image_provider: AsyncImageProvider) -> bool:

{oagi_core-0.9.0 → oagi_core-0.9.1}/src/oagi/agent/tasker/tasker_agent.py

@@ -9,7 +9,7 @@
 import logging
 from typing import Any
 
-from oagi.types import AsyncActionHandler, AsyncImageProvider
+from oagi.types import AsyncActionHandler, AsyncImageProvider, AsyncStepObserver
 
 from ..protocol import AsyncAgent
 from .memory import PlannerMemory
@@ -39,6 +39,7 @@ class TaskerAgent(AsyncAgent):
         temperature: float = 0.0,
         reflection_interval: int = 20,
         planner: Planner | None = None,
+        step_observer: AsyncStepObserver | None = None,
     ):
         """Initialize the tasker agent.
 
@@ -50,6 +51,7 @@ class TaskerAgent(AsyncAgent):
             temperature: Sampling temperature
             reflection_interval: Actions before reflection
             planner: Planner for planning and reflection
+            step_observer: Optional observer for step tracking
         """
         self.api_key = api_key
         self.base_url = base_url
@@ -58,6 +60,7 @@ class TaskerAgent(AsyncAgent):
         self.temperature = temperature
         self.reflection_interval = reflection_interval
         self.planner = planner or Planner()
+        self.step_observer = step_observer
 
         # Memory for tracking workflow
         self.memory = PlannerMemory()
@@ -174,6 +177,7 @@ class TaskerAgent(AsyncAgent):
             planner=self.planner,
             external_memory=self.memory,  # Share memory with child
             todo_index=todo_index,  # Pass the todo index
+            step_observer=self.step_observer,  # Pass step observer
         )
 
         self.current_todo_index = todo_index