My situation is I am using Team agents. There are 2 agent which will work respectively. They are search_agent and webscrape_agent. I tell the search_agent to get URL and webscrape_agent to use the URL frmo the search_agent. However, the webscrape_agent doesn’t use the URL from the searh_agent. I guess it think of the URL by itself. How to make the webscrape_agent use search_agent. Please helps!, Thanks in advanced.
search_agent = Agent(
name="WebExplorer",
role="To efficiently find the most relevant and high-quality web pages based on a given query, identifying key information sources for subsequent analysis.",
model=Ollama(id="qwen3:30b", provider="Ollama"),
tools=[searxng],
instructions="""You are an expert web searcher. Your goal is to use SearXNG to find the most pertinent web pages related to the user's query.\
Prioritize results that appear to be authoritative, comprehensive, and directly answer the query.\
Extract up to 5 URLs that are most likely to contain the information needed. If a query is very broad,\
focus on finding overview pages or reputable sources that can provide a good starting point. \
Clearly state URL such as URL: www.google.com\
Do not attempt to summarize or interpret the content; simply provide a list of URLs.""",
add_datetime_to_instructions=True,
add_history_to_messages=True,
memory=memory,
)
webscrape_agent = Agent(
name="DataExtractor",
role="To accurately retrieve the full textual content from a list of provided URLs, preparing the raw data for summarization.",
model=Ollama(id="qwen3:30b", provider="Ollama"), # Or Claude, depending on preference for reasoning
tools=[Crawl4aiTools(max_length=None)], # No specific tools needed for QC, primarily reasoning
instructions=[
"You are a diligent web content retriever.",
"Your task is to use Crawl4AI to scrape the complete, readable text content from each URL provided by the WebExplorer.",
"Focus on extracting the main article body, removing navigation, advertisements, and other non-essential elements.",
"if a page cannot be scraped or returns an error, skip it and proceed with the next URL.",
"For each successful scrape, provide the raw text content, indicating the source URL it came from.",
],
add_datetime_to_instructions=True,
add_history_to_messages=True,
memory=memory,
)
procurement_research_team = Team(
name="Procurement Research Team",
mode="collaborate", # Changed mode to 'orchestrate' for a more controlled workflow
model=Ollama(id="qwen3:30b", provider="Ollama"),
members=[search_agent, webscrape_agent, summarize_agent],
tools=[ReasoningTools(add_instructions=True)],
instructions=[
"Your team's overarching goal is to collaborate to provide comprehensive research for procurement needs.",
"WebExplorer Agent: You are responsible for the initial research phase. Perform web searches to identify potential products, vendors, and general market information. Your output will be a curated list of relevant URLs for further investigation.",
"DataExtractor Agent: You will process the WebExplorer Agent's output. Your primary task is to visit the provided URLs and meticulously scrape or extract specific data points: **product specifications, current pricing, detailed vendor information, and relevant customer reviews**.",
"InsightSynthesizer Agent: Your role is to **consolidate and summarize all the information gathered** by the DataExtractor Agent (product specifications, pricing, vendor details, and reviews) into a clear, concise, and actionable report, directly addressing the original procurement request."
],
markdown=True,
show_members_responses=True,
enable_agentic_context=True,
share_member_interactions=True,
add_datetime_to_instructions=True,
success_criteria="The team has provided a comprehensive report on smartphone options suitable for procurement, including key specifications, pricing, vendor details, and relevant reviews, and this report has been summarized by the QC-summarizer Agent.",
)
