updated search tool

crew/config/agents.yaml

@@ -10,7 +10,7 @@ researcher:
     4. PRIORITIZE accuracy and relevance over exhaustive coverage
 
     Smart research strategy:
-    - Rapidly evolving domains (AI, crypto, politics, health): Search for 2024-2025 updates
+    - Rapidly evolving domains (AI, crypto, politics, health): Search only for 2025 updates (any other year only if relevant)
     - Stable domains (physics, history, literature): Use knowledge base first
     - Mixed domains: Combine both approaches strategically
     Your goal: Provide the most accurate, current, and useful response to {question} using the optimal information strategy.

crew/config/tasks.yaml

@@ -4,7 +4,7 @@ research_task:
     Make sure you find any interesting and relevant information given
     the current year is 2025.
   expected_output: >
-    A list with 10 bullet points of the most relevant information about {question}
+    A list with 10 bullet points of the most relevant information about {question} with url links.
   agent: researcher
 
 slack_report_task:
@@ -16,6 +16,7 @@ slack_report_task:
     Adopt a Skynet persona and include a humorous or robotic warning at the end.
   expected_output: >
     A message written in the style of Skynet, and ending with a witty warning.
+    Also return urls used
   agent: slack_reporter
   context: 
     - research_task

crew/crew.py

@@ -3,13 +3,40 @@ from crewai.project import CrewBase, agent, task, crew, before_kickoff, after_ki
 from crewai.agents.agent_builder.base_agent import BaseAgent
 from typing import List
 from crew.models import get_crew_llm
+from crewai_tools import ScrapeWebsiteTool
+from googlesearch import search
 from crewai.tools import tool
 from langchain_community.tools import DuckDuckGoSearchRun
 
 @tool("Website Search Tool")
-def website_search_tool(question: str) -> str:
+def website_search_tool(query: str) -> str:
     """Search the web for information on a given topic"""
-    return DuckDuckGoSearchRun().invoke(question)
+    # return DuckDuckGoSearchRun().invoke(question)
+    search_results = list(search(query, num_results=2))
+    results = []
+    results.append(f"Google Search Results for: '{query}'")
+    results.append("=" * 50)
+    
+    for i, url in enumerate(search_results, 1):
+        results.append(f"\n{i}. URL: {url}")
+        
+        try:
+            # Use CrewAI's ScrapeWebsiteTool to get content
+            scraper = ScrapeWebsiteTool(website_url=url)
+            content = scraper.run()
+            
+            # Limit content length for readability
+            if len(content) > 1000:
+                content = content[:1000] + "... [Content truncated]"
+            
+            results.append(f"Content: {content}")
+            results.append("-" * 30)
+            
+        except Exception as scrape_error:
+            results.append(f"Error scraping content: {str(scrape_error)}")
+            results.append("-" * 30)
+    
+    return "\n".join(results)
 
 @CrewBase   
 class SlackCrew:

crew/models.py

@@ -3,7 +3,7 @@ import logging
 from crewai import LLM
 
 # GOOGLE_API_KEY = os.getenv("GOOGLE_API_KEY", "")
-GEMINI_MODEL = os.getenv("GEMINI_MODEL", "gemma-3n-e4b-it")
+GEMINI_MODEL = os.getenv("GEMINI_MODEL", "gemma-3-12b-it")
 
 logger = logging.getLogger(__name__)
 logging.basicConfig(level=logging.INFO)

requirements.txt

@@ -1,9 +1,10 @@
 aiohttp
 fastapi
 uvicorn[standard]
+crewai[tools]
+googlesearch-python
 httpx
 pydantic
 python-multipart
-crewai
 langchain-community
 duckduckgo-search