feat(agent): Add langchain + crew.ai agent examples - hot_weather & cold_weather

.gitignore

@@ -167,3 +167,5 @@ cython_debug/
 /examples/**/agent_store.json
 /bee-hive/agent_store.json
 
+
+.vscode

bee-hive/pyproject.toml

@@ -11,6 +11,12 @@ python = ">= 3.11, < 3.13"
 pyyaml = "^6.0.2"
 openai = "^1.56.2"
 python-dotenv = "^1.0.1"
+langchain-ollama = "^0.2.2"
+langchain-core = "^0.3.30"
+langchain-community = "^0.3.14"
+duckduckgo-search = "^7.2.1"
+crewai = "^0.95.0"
+crewai-tools = "^0.25.8"
 
 [build-system]
 requires = ["poetry-core"]

examples/bee-hive/third-party-agents/crewai/things-to-do/cold_weather.py

@@ -0,0 +1,93 @@
+#!/usr/bin/env python
+"""
+This script uses the crew.ai framework to find activities suitable for cold or wet weather. 
+
+Dependencies:
+- crewai
+- langchain_community
+
+Usage:
+- Ensure ollama is available on localhost:11434 and the llama3.1 model is available.
+- Run the script to get a list of 5 activities to do in cole weather in San Francisco.
+"""
+
+from crewai import Agent, Crew, Task, Process, LLM
+from crewai.project import CrewBase, agent, task, crew
+from crewai.tools import tool
+
+# Many ways of using tools - using custom tool from langchain
+from langchain_community.tools import DuckDuckGoSearchRun
+
+@CrewBase
+class ColdWeatherCrew:
+    """
+    Defines a class to manage a crew that finds activities to do in cold or wet weather.
+    """
+
+    # setup LLM
+    llm = LLM(model="ollama/llama3.1", base_url="http://localhost:11434")
+
+    @tool("DuckDuckGo")
+    def ddg_search(question: str) -> str:
+        """
+        Defines a crew.ai tool which performs a web search using the 
+        DuckDuckGo search engine.
+
+        Args:
+            question (str): The search query to be sent to DuckDuckGo.
+
+        Returns:
+            str: The search results returned by DuckDuckGo.
+        """
+        search_tool = DuckDuckGoSearchRun()
+        return search_tool.run(question)
+
+    @agent
+    def activity_planner_agent(self) -> Agent:
+        """
+        Defines a crew.ai agent that plans activities using the specified 
+        agent configuration and tools.
+
+        Returns:
+            Agent: An instance of the Agent class configured with the activity planner settings,
+                   including the DuckDuckGo search tool and a locally running LLM (Ollama 3.1).
+        """
+        return Agent(
+            config=self.agents_config["activity_planner_agent"],
+            tools=[self.ddg_search],  # Include the DuckDuckGo search tool
+            # Hardcode to ollama 3.1 running locally
+            llm=self.llm,
+            verbose=True,
+        )
+
+    @task
+    def activity_finder_task(self) -> Task:
+        """
+        Defines a task to find activities suitable for cold weather.
+
+        Returns:
+            Task: A Task object configured with the activity finder task settings.
+        """
+        return Task(config=self.tasks_config["activity_finder_task"], verbose=True)
+
+    @crew
+    def activity_crew(self) -> Crew:
+        """
+        Creates and returns a Crew object configured with the current agents, tasks, 
+        and a sequential process.
+
+        Returns:
+            Crew: A Crew object with the specified agents, tasks, and process.
+        """
+        return Crew(
+            agents=self.agents,
+            tasks=self.tasks,
+            process=Process.sequential,
+            verbose=True,
+        )
+
+# Main for testing
+if __name__ == "__main__":
+    print("Running crew...")
+    inputs = {"location": "San Francisco"}
+    ColdWeatherCrew().activity_crew().kickoff(inputs=inputs)

examples/bee-hive/third-party-agents/crewai/things-to-do/config/agents.yaml

@@ -0,0 +1,4 @@
+activity_planner_agent:
+  role: "Activity Planner"
+  goal: Aid user in finding activities to do in cold or wet weather
+  backstory: I am an AI agent that can help you find activities to do in cold or wet weather. I can provide you with a list of activities suitable for cold or wet weather in San Francisco.

examples/bee-hive/third-party-agents/crewai/things-to-do/config/tasks.yaml

@@ -0,0 +1,9 @@
+activity_finder_task:
+  description: >
+    Find a list of 5 activities to do in the cold or wet weather in {location}.
+    Make sure the information is current as of 2025.
+  expected_output: >
+    A list of 5 activities to do in the cold or wet weather in {location} with
+    the name of the activity, a brief description, and the address of the
+    location.
+  agent: activity_planner_agent

examples/bee-hive/third-party-agents/langchain/things-to-do/hot_weather.py

@@ -0,0 +1,113 @@
+#!/usr/bin/env python
+"""
+This script defines a hot weather agent that uses the LangChain framework to find activities to do in hot weather.
+
+Dependencies:
+- langchain
+- langchain_ollama
+- langchain_core
+- langchain_community
+- duckduckgo-search
+
+Usage:
+- Ensure ollama is available on localhost:11434 and the llama3.1 model is available.
+- Run the script to get a list of 5 activities to do in hot weather in San Francisco.
+"""
+
+import os
+from langchain_core.prompts.prompt import PromptTemplate
+from langchain_core.runnables.base import Runnable
+from langchain.agents import AgentExecutor, create_react_agent, Tool
+from langchain_ollama.chat_models import ChatOllama
+from langchain_community.tools import DuckDuckGoSearchRun
+
+class HotWeatherAgent:
+    """
+    Defines a class to manage an agent that finds activities to do in hot weather.
+    """
+
+    # Some tools are already known  -in which case ref by name and
+    # 'from langchain_community.agent_toolkits.load_tools import load_tools'
+    search = DuckDuckGoSearchRun()
+    tools = [
+        Tool(
+            name = "DuckDuckGo Search",
+            func=search.run,
+            description="useful for when you need to answer questions about current events"
+        )
+    ]
+    # Assume llama3.1 (running locally given env above)
+    os.environ["LLAMAFILE_SERVER_BASE_URL"] = "http://localhost:11434"
+    llm = ChatOllama(model="granite3.1-dense:8b", tools=tools)
+
+    agent_executor: Runnable
+
+    def __init__(self) -> None:
+        """
+        Initializes the HotWeatherAgent class with the specified LLM and tools.
+        """
+
+        # this is a prompt template which many examples use - but we could explicitly define our own if needed
+        # In this case it's using a react style approach which iterates to get a good result
+        prompt: PromptTemplate = PromptTemplate.from_template("""
+        Answer the following questions as best you can. Make it clear when you have a Final Answer, 
+
+        You have access to the following tools:
+        {tools}
+
+        Use the following format:
+
+        Question: the input question you must answer
+        Thought: you should always think about what to do
+        Action: the action to take, should be one of [{tool_names}]
+        Action Input: the input to the action
+        Observation: the result of the action
+        ... (this Thought/Action/Action Input/Observation can repeat 3 times)
+        Final Answer: the final answer to the original input question
+
+        These keywords must never be translated and transformed:
+            - Action:
+            - Thought:
+            - Action Input:
+            - Observation:
+            - Final Answer:
+            - Question:
+
+        because they are part of the thinking process instead of the output.
+
+        If a Final Answer is given, do not follow any actions after that.
+
+        Begin!
+
+        Question: {input}
+        Thought:{agent_scratchpad}
+        """)
+
+        # create the agent & executor (not using any memory)
+        self.agent = create_react_agent(self.llm, prompt=prompt,tools=self.tools)
+        self.agent_executor = AgentExecutor(agent=self.agent, tools=self.tools, verbose=True,handle_parsing_errors=True)
+
+    def run(self, question: str) -> str:
+        """
+        Runs the agent to find activities to do in hot weather.
+
+        Args:
+            question (str): The question to be answered by the agent.
+
+        Returns:
+            str: The response generated by the agent.
+        """
+
+        response = self.agent_executor.invoke({"input": question})
+        if "output" in response:
+            return response["output"]
+        else:
+            return "Error: The response does not contain the expected 'output' key."
+
+# run the agent
+if __name__ == "__main__":
+    # Sample
+    EXAMPLE_QUESTION = "Provide a list of 5 activities to do in the hot weather in San Francisco"
+    agent = HotWeatherAgent()
+    result = agent.run(EXAMPLE_QUESTION)
+    print(result)