droidrun 0.5.0.dev5__tar.gz → 0.5.0.dev6__tar.gz

{droidrun-0.5.0.dev5 → droidrun-0.5.0.dev6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: droidrun
-Version: 0.5.0.dev5
+Version: 0.5.0.dev6
 Summary: A framework for controlling Android devices through LLM agents
 Project-URL: Homepage, https://github.com/droidrun/droidrun
 Project-URL: Bug Tracker, https://github.com/droidrun/droidrun/issues

{droidrun-0.5.0.dev5 → droidrun-0.5.0.dev6}/droidrun/agent/codeact/tools_agent.py

@@ -10,6 +10,7 @@ compatibility with DroidAgent's execute_task() method.
 
 import asyncio
 import logging
+import os
 from typing import TYPE_CHECKING, Optional, Type
 
 from llama_index.core.llms.llm import LLM
@@ -88,6 +89,9 @@ class FastAgent(Workflow):
         self.action_ctx = action_ctx
         self.state_provider = state_provider
         self.save_trajectory = save_trajectory
+        self._stream_screenshots = os.environ.get(
+            "DROIDRUN_STREAM_SCREENSHOTS", ""
+        ).lower() in ("1", "true")
         self.shared_state = shared_state
         self.output_model = output_model
         self.prompt_resolver = prompt_resolver or PromptResolver()
@@ -218,7 +222,7 @@ class FastAgent(Workflow):
 
         # Capture screenshot if needed
         screenshot = None
-        if self.vision or self.save_trajectory != "none":
+        if self.vision or self._stream_screenshots or self.save_trajectory != "none":
             try:
                 screenshot = await self.action_ctx.driver.screenshot()
 

{droidrun-0.5.0.dev5 → droidrun-0.5.0.dev6}/droidrun/agent/droid/droid_agent.py

@@ -8,6 +8,7 @@ Architecture:
 """
 
 import logging
+import os
 import traceback
 from typing import TYPE_CHECKING, Awaitable, Type, Union
 
@@ -208,6 +209,10 @@ class DroidAgent(Workflow):
         # Check if using external agent - skip LLM loading
         self._using_external_agent = self.config.agent.name != "droidrun"
 
+        self._stream_screenshots = os.environ.get(
+            "DROIDRUN_STREAM_SCREENSHOTS", ""
+        ).lower() in ("1", "true")
+
         self.timeout = timeout
 
         # Store user custom tools
@@ -1039,17 +1044,21 @@ class DroidAgent(Workflow):
                 if self.config.logging.debug:
                     logger.error(traceback.format_exc())
 
-        # Capture final screenshot before saving trajectory
-        if self.config.logging.save_trajectory != "none":
+        # Capture final screenshot (independent of trajectory persistence)
+        vision_any = (
+            self.config.agent.manager.vision
+            or self.config.agent.executor.vision
+            or self.config.agent.fast_agent.vision
+        )
+        if (
+            vision_any
+            or self._stream_screenshots
+            or self.config.logging.save_trajectory != "none"
+        ):
             try:
                 screenshot = await self.action_ctx.driver.screenshot()
                 if screenshot:
                     ctx.write_event_to_stream(ScreenshotEvent(screenshot=screenshot))
-                    vision_any = (
-                        self.config.agent.manager.vision
-                        or self.config.agent.executor.vision
-                        or self.config.agent.fast_agent.vision
-                    )
                     parent_span = trace.get_current_span()
                     record_langfuse_screenshot(
                         screenshot,
@@ -1061,6 +1070,8 @@ class DroidAgent(Workflow):
             except Exception as e:
                 logger.warning(f"Failed to capture final screenshot: {e}")
 
+        # Save trajectory to disk
+        if self.config.logging.save_trajectory != "none":
             # Populate macro data from RecordingDriver log
             if isinstance(self.driver, RecordingDriver):
                 self.trajectory.macro = list(self.driver.log)

{droidrun-0.5.0.dev5 → droidrun-0.5.0.dev6}/droidrun/agent/manager/manager_agent.py

@@ -13,6 +13,7 @@ from __future__ import annotations
 import copy
 import json
 import logging
+import os
 from typing import TYPE_CHECKING, Optional, Type
 
 from llama_index.core.llms.llm import LLM
@@ -88,6 +89,9 @@ class ManagerAgent(Workflow):
         self.action_ctx = action_ctx
         self.state_provider = state_provider
         self.save_trajectory = save_trajectory
+        self._stream_screenshots = os.environ.get(
+            "DROIDRUN_STREAM_SCREENSHOTS", ""
+        ).lower() in ("1", "true")
         self.shared_state = shared_state
         self.registry = registry
         self.output_model = output_model
@@ -412,7 +416,7 @@ class ManagerAgent(Workflow):
 
         # Capture screenshot if needed
         screenshot = None
-        if self.vision or self.save_trajectory != "none":
+        if self.vision or self._stream_screenshots or self.save_trajectory != "none":
             try:
                 screenshot = await self.action_ctx.driver.screenshot()
 

{droidrun-0.5.0.dev5 → droidrun-0.5.0.dev6}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "droidrun"
-version = "0.5.0.dev5"
+version = "0.5.0.dev6"
 description = "A framework for controlling Android devices through LLM agents"
 authors = [{ name = "Niels Schmidt", email = "niels.schmidt@droidrun.ai" }]
 dependencies = [

{droidrun-0.5.0.dev5 → droidrun-0.5.0.dev6}/uv.lock

@@ -760,7 +760,7 @@ wheels = [
 
 [[package]]
 name = "droidrun"
-version = "0.5.0.dev5"
+version = "0.5.0.dev6"
 source = { editable = "." }
 dependencies = [
     { name = "aiofiles" },