new

agent/manager_agent.py

@@ -19,7 +19,7 @@ class ManagerAgent(AgentBase):
                 call_openai_res = json.loads(self.call_gpt(prompt))
                 if call_openai_res["isOk"] == "no":
                     okr_object.raw_user_task = okr_object.raw_user_task + f"，{i['target']}：" + input(
-                        f"【警告】{call_openai_res['content']}\n请您补充信息：")
+                        f"【任务】{okr_object.raw_user_task}\n\n【警告】{call_openai_res['content']}\n请您补充信息：")
             i['content'] = call_openai_res["content"]
 
 

agent/plan_agent.py

@@ -6,7 +6,7 @@ from work_principle.okr_principle import OKR_Object
 class PlanAgent(AgentBase):
     def __init__(self):
         self.logger = logging.getLogger(__name__)
-        super().__init__('你是一名计划拆解者，利用工具将object拆解成key_result，请你返回json格式的内容，比如["使用工具a进行xx操作", "使用工具b进行xx操作"]')
+        super().__init__('你是一名计划拆解者，利用工具将object拆解成key_result，请你返回json格式的内容，比如[{"tools_name":"web_browser","param":"工具a"}]')
 
     # 根据object和已有的工具能力拆解成key result，key
     def aligning(self, okr_object: OKR_Object):

tools/llm_tools.py

@@ -6,7 +6,9 @@ from tools.tools_base import ToolsBase
 class LLMTools(ToolsBase):
     def __init__(self):
         self.name = "llm_tools"
-        self.description = "利用大模型进行回答，入参格式为字典：{role:大模型的角色, content:问题}"
+        self.description = "利用大模型进行回答"
+        self.request_param='字典，如{"content": "天气怎么样"}'
+        self.return_content = "大模型结果"
     
     def run(self, param=None):
         """

tools/tools_base.py

@@ -2,9 +2,11 @@ class ToolsBase:
     def __init__(self):
         self.name = ""
         self.description = ""
+        self.request_param = ""
+        self.return_content = ""
         
     def get_info(self):
-        return {"name": self.name, "description": self.description}
+        return {"name": self.name, "description": self.description, "param": self.request_param, "return_content": self.return_content}
 
     def run(self, param=None):
         pass

tools/web_browser_element_tools.py

@@ -0,0 +1,31 @@
+import os
+import yaml
+from selenium import webdriver
+from selenium.webdriver.common.by import By
+from webdriver_manager.chrome import ChromeDriverManager
+from webdriver_manager.firefox import GeckoDriverManager
+from webdriver_manager.microsoft import EdgeChromiumDriverManager
+from webdriver_manager.opera import OperaDriverManager
+from selenium.webdriver.edge.service import Service as EdgeService
+from selenium.webdriver.common.by import By
+
+from tools.tools_base import ToolsBase
+
+class WebBrowserElement(ToolsBase):
+    def __init__(self):
+        self.name = "web_element"
+        self.description = "利用selenium对指定xpath进行点击或者输入内容"
+        self.request_param = '字典，如{"driver": "driver", "xpath": "", "action":"click或send_text"}'
+        self.return_content = "True"
+
+    def run(self, param=None):
+        driver = param["driver"]
+        xpath = param["xpath"]
+        action = param["action"]
+        if action == "click":
+            driver.find_element(By.XPATH, xpath).click()
+        elif action == "send_text":
+            driver.find_element(By.XPATH, xpath).send_keys(param["text"])
+        return True
+
+       

tools/web_browser_url_tools.py

@@ -11,10 +11,12 @@ from selenium.webdriver.common.by import By
 
 from tools.tools_base import ToolsBase
 
-class WebBrowser(ToolsBase):
+class WebBrowserUrl(ToolsBase):
     def __init__(self):
         self.name = "web_browser"
-        self.description = "利用浏览器进行搜索，入参格式为字符串"
+        self.description = "利用selenium对指定URL进行访问"
+        self.request_param = '字典，如{"usrl": ""}'
+        self.return_content = '{"driver": "selenium的webdriver，driver用于继续在此见面上进行操作，可作为web_element工具的入参", "content": "网页xml结构"'
 
     def run(self, param=None):
         # Load browser configuration from YAML file
@@ -47,9 +49,8 @@ class WebBrowser(ToolsBase):
                 webdriver_manager = OperaDriverManager()
                 driver = webdriver.Opera(webdriver_manager.install(), options=options)
         driver.implicitly_wait(10)
-        browser_url = config.get("browser_url")
-        driver.get(browser_url)
-        driver.find_element(by=By.XPATH, value='//*[@id="kw"]').send_keys("陈")
-        driver.find_element(by=By.XPATH, value='//*[@id="su"]').click()
-        print(driver.page_source)
+        # browser_url = config.get("browser_url")
+        # driver.get(browser_url)
+        driver.get(param["url"])
         driver.quit()
+        return driver.page_source