updates

agent.py

@@ -1,6 +1,7 @@
 import os
+from typing import Annotated
 
-from langchain_community.document_loaders import WikipediaLoader
+from langchain_community.document_loaders import WikipediaLoader, YouTubeLoader
 
 # --- Langchain / Langraph ---
 from langchain_community.tools.tavily_search import TavilySearchResults
@@ -19,28 +20,20 @@ from langgraph.graph.message import add_messages
 from langgraph.prebuilt import ToolNode, tools_condition
 
 
-@tool
-def search_web_sources(query: str) -> dict:
-    """
-    Perform a web search using Tavily and return up to 3 relevant documents.
-    This tool is useful for answering research-based queries that require
-    up-to-date information from trusted sources.
+@tool("search_web_sources")
+def search_web_sources(query: Annotated[str, "Search query string"]) -> dict:
+    """Performs a web search and returns up to 3 formatted documents with content and source."""
 
-    Args:
-        query (str): The input search query.
-
-    Returns:
-        str: Formatted web search results with metadata and content.
-    """
-    tavily_tool = TavilySearchResults(max_results=3)
-    search_docs = tavily_tool.invoke({"query": query})
-    formatted_search_docs = "\n\n---\n\n".join(
+    if not os.environ.get("TAVILY_API_KEY"):
+        raise EnvironmentError("TAVILY_API_KEY is not set in environment variables.")
+    search_docs = TavilySearchResults(max_results=3).invoke({"query": query})
+    formatted = "\n\n---\n\n".join(
         [
-            f'<Document source="{doc.metadata["source"]}" page="{doc.metadata.get("page", "")}"/>\n{doc.page_content}\n</Document>'
+            f'<Document source="{doc.metadata["source"]}" page="{doc.metadata.get("page", "")}">\n{doc.page_content}\n</Document>'
             for doc in search_docs
         ]
     )
-    return {"web_results": formatted_search_docs}
+    return {"web_results": formatted}
 
 
 @tool
@@ -63,6 +56,26 @@ def search_wikipedia(query: str) -> dict:
         return {"wiki_results": f"Error fetching Wikipedia article: {e}"}
 
 
+@tool
+def extract_youtube_transcript(video_url: str) -> dict:
+    """Extract transcript from a YouTube video given its URL using LangChain's YouTubeLoader."""
+    try:
+        loader = YouTubeLoader(video_urls=[video_url])
+        docs = loader.load()
+        if docs:
+            formatted_docs = "\n\n---\n\n".join(
+                [
+                    f'<YouTubeTranscript url="{video_url}">\n{doc.page_content}\n</YouTubeTranscript>'
+                    for doc in docs
+                ]
+            )
+            return {"transcript_results": formatted_docs}
+        else:
+            return {"transcript_results": "No transcript found."}
+    except Exception as e:
+        return {"transcript_results": f"Error fetching YouTube transcript: {e}"}
+
+
 @tool
 def run_python_code(code: str) -> str:
     """Execute Python code and return the result.
@@ -96,7 +109,12 @@ Your response must always begin with: FINAL ANSWER:
 def build_agent_graph(provider: str = "groq"):
 
     # Define toolset
-    tools = [search_web_sources, search_wikipedia, run_python_code]
+    tools = [
+        search_web_sources,
+        search_wikipedia,
+        extract_youtube_transcript,
+        run_python_code,
+    ]
 
     # Instantiate LLM
     os.environ["GROQ_API_KEY"]

requirements.txt

@@ -9,3 +9,5 @@ langchain-experimental
 langchain-tavily
 langgraph
 tavily-python
+wikipedia
+youtube-transcript-api