Add youtube tool

community_tools.py

@@ -1,6 +1,7 @@
 from langchain_google_community import GooglePlacesTool
 from langchain_community.agent_toolkits.load_tools import load_tools
-from smolagents.tools import Tool
+from langchain_community.document_loaders import YoutubeLoader
+from smolagents.tools import Tool, tool
 
 google_map_tool = Tool.from_langchain(GooglePlacesTool())
 
@@ -12,3 +13,19 @@ wikipedia_tool = Tool.from_langchain(WikipediaQueryRun(api_wrapper=WikipediaAPIW
 arxiv_tools = [Tool.from_langchain(tool) for tool in load_tools(["arxiv"])]
 
 community_tools = [google_map_tool, wikipedia_tool, *arxiv_tools]
+
+@tool
+def get_youtube_transcript_from_url(video_url: str = None):
+    """
+    Get the transcript of a YouTube video
+    Args:
+        video_url: The URL of the YouTube video
+    Returns:
+        The transcript of the YouTube video
+    """
+    youtube_loader = YoutubeLoader(video_url=video_url)
+    docs = youtube_loader.load()
+    transcript = docs[0].page_content
+    return transcript
+
+get_youtube_transcript_from_url_tool = get_youtube_transcript_from_url()

mini_agents.py

@@ -4,7 +4,7 @@ from tools import tavily_search_tool, visit_webpage_tool
 from tools import to_dataframe, to_json, get_dataframe_data, get_dataframe_column, get_dataframe_row, get_dataframe_groupby
 from vlm_tools import download_image, image_processing, object_detection_tool, ocr_scan_tool, extract_frames_from_video
 from audio_tools import transcribe_audio_tool, audio_to_base64, noise_reduction, audio_segmentation, speaker_diarization
-from community_tools import community_tools
+from community_tools import community_tools, get_youtube_transcript_from_url_tool
 import os
 import logging
 import yaml
@@ -124,7 +124,7 @@ master_model = InferenceClientModel(
 master_agent = CodeAgent(
     model=master_model,
     managed_agents=[audio_agent, vlm_agent, arithmetic_agent, pandas_agent],
-    tools=[sort_list, *community_tools, tavily_search_tool, visit_webpage_tool],
+    tools=[sort_list, *community_tools, tavily_search_tool, visit_webpage_tool, get_youtube_transcript_from_url_tool],
     add_base_tools=True,
     max_steps=20,
     additional_authorized_imports=["math", "pandas", "json", "numpy", "io", "os", "logging", "yaml", "pyplot", "matplotlib", 'hmmlearn', 'pickle', 'sklearn', 'scipy', 'datetime', 'typing'],

requirements.txt

@@ -16,6 +16,7 @@ coloredlogs==15.0.1
 contourpy==1.3.2
 cycler==0.12.1
 dataclasses-json==0.6.7
+defusedxml==0.7.1
 deprecation==2.1.0
 duckduckgo_search==8.0.2
 eyeD3==0.9.8
@@ -98,6 +99,7 @@ pytesseract==0.3.13
 python-dateutil==2.9.0.post0
 python-dotenv==1.1.0
 python-multipart==0.0.20
+pytube==15.0.0
 pytz==2025.2
 PyYAML==6.0.2
 requests==2.32.3
@@ -135,4 +137,5 @@ websockets==15.0.1
 wikipedia==1.4.0
 Wikipedia-API==0.8.1
 yarl==1.20.0
+youtube-transcript-api==1.0.3
 zstandard==0.23.0