opensource/Agent-S
3
0
Fork 0
mirror of https://github.com/simular-ai/Agent-S.git synced 2026-09-24 23:09:42 +08:00
Code Issues Packages Projects Releases Wiki Activity

wip readme

Browse Source
This commit is contained in:
alckasoc
2025-03-11 18:14:14 -07:00
parent 473c4cf662
commit f404d9c2bd
3 changed files with 2 additions and 74 deletions
Download Patch File Download Diff File Expand all files Collapse all files
SERVER_README.md
View File
gui_agents/s1/OSWorld.md
+2 -2
View File
@@ -26,8 +26,8 @@ We suggest creating a separate conda environment for each repository to avoid de
After completing the setup instructions, import the GraphSearchAgent into the run.py file in OSWorld. The GraphSearchAgent is the parent agent used in the Agent S framework. To understand the architecture of this GraphSearchAgent, refer to [Agent S Architecture](images/agent_s_architecture.pdf).
```
from gui_agents.aci.LinuxOSACI import LinuxACI
from gui_agents.core.AgentS import GraphSearchAgent
from gui_agents.s1.aci.LinuxOSACI import LinuxACI
from gui_agents.s1.core.AgentS import GraphSearchAgent
```
Replace the PromptAgent on line 138 in the test() method with the Graph Search Agent. Specify engine params and instantiate the agent as shown:
gui_agents/s1/aci/LinuxOSACI.py
-72
View File
@@ -94,8 +94,6 @@ subprocess.run(['wmctrl', '-ir', window_id, '-b', 'add,maximized_vert,maximized_
component_ns = "https://accessibility.ubuntu.example.org/ns/component"
value_ns = "https://accessibility.ubuntu.example.org/ns/value"
self._existing_bash_session = False
def get_active_apps(self, obs: Dict) -> List[str]:
tree = ET.ElementTree(ET.fromstring(obs["accessibility_tree"]))
apps = []
@@ -459,76 +457,6 @@ subprocess.run(['wmctrl', '-ir', window_id, '-b', 'add,maximized_vert,maximized_
return command
# if overwrite:
# return f"""import pyautogui; pyautogui.click({x}, {y}); pyautogui.hotkey("ctrl", "a"); pyautogui.press("backspace"); pyautogui.typewrite({repr(text)})"""
# else:
# return f"""import pyautogui; pyautogui.click({x}, {y}); pyautogui.hotkey("ctrl", "a"); pyautogui.press("backspace"); pyautogui.typewrite("{text}")"""
# @agent_action
# def type_and_enter(self, element_id:int, text:str, overwrite: bool = True):
# '''Type text into the element and press enter
# Args:
# element_id:int ID of the element to type into
# text:str the text to type into the element
# '''
# try:
# node = self.find_element(element_id)
# except:
# node = self.find_element(0)
# # print(node.attrib)
# coordinates = eval(
# node.get("{{{:}}}screencoord".format(component_ns), "(-1, -1)"))
# sizes = eval(node.get("{{{:}}}size".format(component_ns), "(-1, -1)"))
# # Calculate the center of the element
# x = coordinates[0] + sizes[0] // 2
# y = coordinates[1] + sizes[1] // 2
# # Return pyautoguicode to type into the element
# if overwrite:
# return f"""import pyautogui; pyautogui.click({x}, {y}); pyautogui.hotkey("ctrl", "a"); pyautogui.press("backspace"); pyautogui.typewrite({repr(text)}); pyautogui.press("enter")"""
# else:
# return f"""import pyautogui; pyautogui.click({x}, {y}); pyautogui.typewrite({repr(text)}); pyautogui.press("enter")"""
# @agent_action
# def copy_text(self, element_id:int):
# '''Copy the selected text, use instead of ctrl+c
# Args:
# element_id:int ID of the element to copy text from
# '''
# try:
# node = self.find_element(element_id)
# except:
# node = self.find_element(0)
# self.clipboard = node.text
# @agent_action
# def paste_text(self, element_id:int, overwrite: bool = True):
# '''Paste text from the clipboard into the element, use instead of ctrl+v
# Args:
# element_id:int ID of the element to copy text from
# overwrite:bool a boolean value to determine if the text should be pasted over the existing text or appended to it
# '''
# try:
# node = self.find_element(element_id)
# except:
# node = self.find_element(0)
# coordinates = eval(
# node.get("{{{:}}}screencoord".format(component_ns), "(-1, -1)"))
# sizes = eval(node.get("{{{:}}}size".format(component_ns), "(-1, -1)"))
# # Calculate the center of the element
# x = coordinates[0] + sizes[0] // 2
# y = coordinates[1] + sizes[1] // 2
# # Return pyautoguicode to paste into the element
# if overwrite:
# return f"""import pyautogui; pyautogui.click({x}, {y}); pyautogui.typewrite("{self.clipboard}");"""
# else:
# return f"""import pyautogui; pyautogui.click({x}, {y}); pyautogui.hotkey("ctrl", "a"); pyautogui.press("backspace"); pyautogui.typewrite("{self.clipboard}");"""
@agent_action
def save_to_knowledge(self, text: List[str]):
"""Save facts, elements, texts, etc. to a long-term knowledge bank for reuse during this task. Can be used for copy-pasting text, saving elements, etc.