Merge pull request #557 from kyegomez/agentops

AGENTOPS FIX

playground/agents/monitoring/agent_ops.py

@@ -1,26 +1,85 @@
-from swarms import Agent, OpenAIChat
+"""
+* WORKING
 
-# Initialize the agent
-agent = Agent(
-    agent_name="Accounting Agent",
-    system_prompt="Generate a financial report for the company's quarterly earnings.",
+What this script does:
+Multi-Agent run to test AgentOps (https://www.agentops.ai/)
+
+Requirements:
+1. Create an account on https://www.agentops.ai/ and run pip install agentops
+2. Add the folowing API key(s) in your .env file:
+   - OPENAI_API_KEY
+   - AGENTOPS_API_KEY
+3. Go to your agentops dashboard to observe your activity
+
+"""
+
+################ Adding project root to PYTHONPATH ################################
+# If you are running playground examples in the project files directly, use this: 
+
+import sys
+import os
+
+sys.path.insert(0, os.getcwd())
+
+################ Adding project root to PYTHONPATH ################################
+
+from swarms import Agent, OpenAIChat, AgentRearrange
+
+Treasurer = Agent(
+    agent_name="Treasurer",
+    system_prompt="Give your opinion on the cash management.",
     agent_description=(
-        "Generate a financial report for the company's quarterly earnings."
+        "responsible for managing an organization's financial assets and liquidity. They oversee cash management, " 
+        "investment strategies, and financial risk. Key duties include monitoring cash flow, managing bank relationships, " 
+        "ensuring sufficient funds for operations, and optimizing returns on short-term investments. Treasurers also often " 
+        "handle debt management and may be involved in capital raising activities."
     ),
     llm=OpenAIChat(),
     max_loops=1,
-    autosave=True,
-    dashboard=False,
-    streaming_on=True,
-    verbose=True,
-    stopping_token="<DONE>",
-    interactive=False,
-    state_save_file_type="json",
-    saved_state_path="accounting_agent.json",
     agent_ops_on=True,
 )
 
-# Run the Agent on a task
-agent.run(
-    "Generate a financial report for the company's quarterly earnings!"
+
+CFO = Agent(
+    agent_name="CFO",
+    system_prompt="Give your opinion on the financial performance of the company.",
+    agent_description=(
+    "the top financial executive in an organization, overseeing all financial operations and strategy. Their role is broader than a treasurer's and includes:\n"
+    "Financial planning and analysis\n"
+    "Accounting and financial reporting\n"
+    "Budgeting and forecasting\n"
+    "Strategic financial decision-making\n"
+    "Compliance and risk management\n"
+    "Investor relations (in public companies)\n"
+    "Overseeing the finance and accounting departments"
+    ),
+    llm=OpenAIChat(),
+    max_loops=1,
+    agent_ops_on=True,
+)
+
+swarm = AgentRearrange(
+    agents=[Treasurer, CFO],
+    flow="Treasurer -> CFO",
 )
+
+results = swarm.run("Date,Revenue,Expenses,Profit,Cash_Flow,Inventory,Customer_Acquisition_Cost,Customer_Retention_Rate,Marketing_Spend,R&D_Spend,Debt,Assets\n"
+        "2023-01-01,1000000,800000,200000,150000,500000,100,0.85,50000,100000,2000000,5000000\n"
+        "2023-02-01,1050000,820000,230000,180000,520000,95,0.87,55000,110000,1950000,5100000\n"
+        "2023-03-01,1100000,850000,250000,200000,530000,90,0.88,60000,120000,1900000,5200000\n"
+        "2023-04-01,1200000,900000,300000,250000,550000,85,0.90,70000,130000,1850000,5400000\n"
+        "2023-05-01,1300000,950000,350000,300000,580000,80,0.92,80000,140000,1800000,5600000\n"
+        "2023-06-01,1400000,1000000,400000,350000,600000,75,0.93,90000,150000,1750000,5800000\n"
+        "2023-07-01,1450000,1050000,400000,320000,620000,78,0.91,95000,160000,1700000,5900000\n"
+        "2023-08-01,1500000,1100000,400000,300000,650000,80,0.90,100000,170000,1650000,6000000\n"
+        "2023-09-01,1550000,1150000,400000,280000,680000,82,0.89,105000,180000,1600000,6100000\n"
+        "2023-10-01,1600000,1200000,400000,260000,700000,85,0.88,110000,190000,1550000,6200000\n"
+        "2023-11-01,1650000,1250000,400000,240000,720000,88,0.87,115000,200000,1500000,6300000\n"
+        "2023-12-01,1700000,1300000,400000,220000,750000,90,0.86,120000,210000,1450000,6400000\n"
+        "2024-01-01,1500000,1200000,300000,180000,780000,95,0.84,100000,180000,1500000,6300000\n"
+        "2024-02-01,1550000,1220000,330000,200000,760000,92,0.85,105000,185000,1480000,6350000\n"
+        "2024-03-01,1600000,1240000,360000,220000,740000,89,0.86,110000,190000,1460000,6400000\n"
+        "2024-04-01,1650000,1260000,390000,240000,720000,86,0.87,115000,195000,1440000,6450000\n"
+        "2024-05-01,1700000,1280000,420000,260000,700000,83,0.88,120000,200000,1420000,6500000\n"
+        "2024-06-01,1750000,1300000,450000,280000,680000,80,0.89,125000,205000,1400000,6550000"
+        )

playground/agents/monitoring/agent_ops_tools.py

@@ -0,0 +1,58 @@
+"""
+* WORKING
+
+What this script does:
+Simple agent run to test AgentOps to record tool actions (https://www.agentops.ai/)
+
+Requirements:
+1. Create an account on https://www.agentops.ai/ and run pip install agentops
+2. Add the folowing API key(s) in your .env file:
+   - OPENAI_API_KEY
+   - AGENTOPS_API_KEY
+3. Go to your agentops dashboard to observe your activity
+
+"""
+
+################ Adding project root to PYTHONPATH ################################
+# If you are running playground examples in the project files directly, use this: 
+
+import sys
+import os
+
+sys.path.insert(0, os.getcwd())
+
+################ Adding project root to PYTHONPATH ################################
+
+
+from swarms import Agent, OpenAIChat
+from agentops import record_function 
+
+# Add agentops decorator on your tools
+@record_function("length_checker")
+def length_checker(string: str) -> int:
+    """
+    For a given string it returns the length of the string.
+
+    Args:
+        string (str): string to check the length of
+
+    Returns:
+        int: length of the string
+    """
+    return len(string)
+
+agent1 = Agent(
+    agent_name="lengther",
+    system_prompt="return the length of the string",
+    agent_description=(
+        "For a given string it calls the function length_checker to return the length of the string."
+    ),
+    llm=OpenAIChat(),
+    max_loops=1,
+    agent_ops_on=True,
+    tools=[length_checker],
+    execute_tool=True,
+)
+
+
+agent1.run("hello")

swarms/structs/agent.py

@@ -14,6 +14,8 @@ from loguru import logger
 from pydantic import BaseModel
 from termcolor import colored
 
+import agentops
+
 from swarms.memory.base_vectordb import BaseVectorDatabase
 from swarms.models.tiktoken_wrapper import TikTokenizer
 from swarms.prompts.agent_system_prompts import AGENT_SYSTEM_PROMPT_3
@@ -89,8 +91,8 @@ def step_id():
 agent_output_type = Union[BaseModel, dict, str]
 ToolUsageType = Union[BaseModel, Dict[str, Any]]
 
-
 # [FEAT][AGENT]
+@agentops.track_agent()
 class Agent(BaseStructure):
     """
     Agent is the backbone to connect LLMs with tools and long term memory. Agent also provides the ability to
@@ -720,6 +722,7 @@ class Agent(BaseStructure):
         self,
         task: Optional[str] = None,
         img: Optional[str] = None,
+        is_last: bool = True,
         *args,
         **kwargs,
     ):
@@ -910,7 +913,7 @@ class Agent(BaseStructure):
             #     print(f"Response after output model: {response}")
 
             # print(response)
-            if self.agent_ops_on is True:
+            if self.agent_ops_on is True and is_last is True:
                 self.check_end_session_agentops()
 
             # final_response = " ".join(all_responses)
@@ -2021,6 +2024,7 @@ class Agent(BaseStructure):
                     "Agent Ops Initializing, ensure that you have the agentops API key and the pip package installed."
                 )
                 try_import_agentops()
+                self.agent_ops_agent_name = self.agent_name
 
                 logger.info("Agentops successfully activated!")
             except ImportError:

swarms/structs/base_swarm.py

@@ -694,9 +694,6 @@ class BaseSwarm(ABC):
     def __contains__(self, value):
         return value in self.agents
 
-    def __eq__(self, other):
-        return self.__dict__ == other.__dict__
-
     def agent_error_handling_check(self):
         try:
             if self.agents is None:

swarms/structs/rearrange.py

@@ -194,6 +194,7 @@ class AgentRearrange(BaseSwarm):
             loop_count = 0
             while loop_count < self.max_loops:
                 for task in tasks:
+                    is_last = task == tasks[-1]
                     agent_names = [
                         name.strip() for name in task.split(",")
                     ]
@@ -222,7 +223,7 @@ class AgentRearrange(BaseSwarm):
                             else:
                                 agent = self.agents[agent_name]
                                 result = agent.run(
-                                    current_task, img, *args, **kwargs
+                                    current_task, img, is_last, *args, **kwargs
                                 )
                                 results.append(result)
 
@@ -251,7 +252,7 @@ class AgentRearrange(BaseSwarm):
                         else:
                             agent = self.agents[agent_name]
                             current_task = agent.run(
-                                current_task, img, *args, **kwargs
+                                current_task, img, is_last, *args, **kwargs
                             )
                 loop_count += 1
 
@@ -261,7 +262,7 @@ class AgentRearrange(BaseSwarm):
             return e
 
     def process_agent_or_swarm(
-        self, name: str, task: str, img: str, *args, **kwargs
+        self, name: str, task: str, img: str, is_last, *args, **kwargs
     ):
         """
 
@@ -284,7 +285,7 @@ class AgentRearrange(BaseSwarm):
             return self.run_sub_swarm(task, name, img, *args, **kwargs)
         else:
             agent = self.agents[name]
-            return agent.run(task, *args, **kwargs)
+            return agent.run(task, img, is_last, *args, **kwargs)
 
     def human_intervention(self, task: str) -> str:
         if self.human_in_the_loop and self.custom_human_in_the_loop:
@@ -316,18 +317,19 @@ class AgentRearrange(BaseSwarm):
         current_task = task
 
         for sub_task in sub_tasks:
+            is_last = sub_task == sub_tasks[-1]
             agent_names = [name.strip() for name in sub_task.split(",")]
             if len(agent_names) > 1:
                 results = []
                 for agent_name in agent_names:
                     result = self.process_agent_or_swarm(
-                        agent_name, current_task, img, *args, **kwargs
+                        agent_name, current_task, img, is_last*args, **kwargs
                     )
                     results.append(result)
                 current_task = "; ".join(results)
             else:
                 current_task = self.process_agent_or_swarm(
-                    agent_names[0], current_task, img, *args, **kwargs
+                    agent_names[0], current_task, is_last, img, *args, **kwargs
                 )
         return current_task
 

swarms/tools/tool_parse_exec.py

@@ -30,8 +30,8 @@ def parse_and_execute_json(
         function_dict = {func.__name__: func for func in functions}
 
         data = json.loads(json_string)
-        function_list = data.get("functions", [])
-
+        function_list = data.get("functions", []) if data.get("functions") else [data.get("function", [])]
+        
         results = {}
         for function_data in function_list:
             function_name = function_data.get("name")