Add `YoutubeTranscriptReader` tool integration, update dependencies, and enhance agent functionality.

README.md

@@ -34,6 +34,7 @@ Tools :
 - [x] Wikipedia
 - [x] Image queries
 - [ ] Video queries
+- [x] Video audio transcripts
 - [x] Website parser
 - [x] Text chunker ==> via LoadAndSearchToolSpec
 

pyproject.toml

@@ -14,6 +14,7 @@ dependencies = [
     "llama-index-multi-modal-llms-mistralai>=0.4.0",
     "llama-index-multi-modal-llms-nebius>=0.4.0",
     "llama-index-readers-web>=0.4.2",
+    "llama-index-readers-youtube-transcript>=0.3.0",
     "llama-index-tools-duckduckgo>=0.3.0",
     "llama-index-tools-requests>=0.4.0",
     "llama-index-tools-wikipedia>=0.3.0",

src/gaia_solving_agent/tools.py

@@ -6,6 +6,7 @@ from llama_index.core.tools.tool_spec.load_and_search import LoadAndSearchToolSp
 from llama_index.multi_modal_llms.mistralai import MistralAIMultiModal
 from llama_index.multi_modal_llms.nebius import NebiusMultiModal
 from llama_index.readers.web import SimpleWebPageReader
+from llama_index.readers.youtube_transcript import YoutubeTranscriptReader
 from tavily import AsyncTavilyClient
 
 from gaia_solving_agent import TAVILY_API_KEY, NEBIUS_API_KEY, MISTRAL_API_KEY
@@ -62,7 +63,14 @@ async def vllm_ask_image(query: str, images: ImageDocument | list[ImageDocument]
 
 simple_web_page_reader_tool = OnDemandLoaderTool.from_defaults(
     SimpleWebPageReader(html_to_text=True),
-    name="Tool for loading content from a web page and return it as text",
-    description="A tool for loading data and querying articles from Wikipedia",
+    name="simple_web_page_reader_tool",
+    description="Tool for loading content from a web page and return it as text",
 )
-simple_web_page_reader_toolspec = LoadAndSearchToolSpec.from_defaults(simple_web_page_reader_tool)
+simple_web_page_reader_toolspec = LoadAndSearchToolSpec.from_defaults(simple_web_page_reader_tool)
+
+youtube_transcript_reader_tool = OnDemandLoaderTool.from_defaults(
+    YoutubeTranscriptReader(),
+    name="youtube_transcript_reader_tool",
+    description="Tool for loading the audio transcript from a youtube video and return it as text",
+)
+youtube_transcript_reader_toolspec = LoadAndSearchToolSpec.from_defaults(youtube_transcript_reader_tool)

uv.lock

@@ -39,6 +39,7 @@ dependencies = [
     { name = "llama-index-multi-modal-llms-mistralai" },
     { name = "llama-index-multi-modal-llms-nebius" },
     { name = "llama-index-readers-web" },
+    { name = "llama-index-readers-youtube-transcript" },
     { name = "llama-index-tools-duckduckgo" },
     { name = "llama-index-tools-requests" },
     { name = "llama-index-tools-wikipedia" },
@@ -67,6 +68,7 @@ requires-dist = [
     { name = "llama-index-multi-modal-llms-mistralai", specifier = ">=0.4.0" },
     { name = "llama-index-multi-modal-llms-nebius", specifier = ">=0.4.0" },
     { name = "llama-index-readers-web", specifier = ">=0.4.2" },
+    { name = "llama-index-readers-youtube-transcript", specifier = ">=0.3.0" },
     { name = "llama-index-tools-duckduckgo", specifier = ">=0.3.0" },
     { name = "llama-index-tools-requests", specifier = ">=0.4.0" },
     { name = "llama-index-tools-wikipedia", specifier = ">=0.3.0" },
@@ -1755,6 +1757,19 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/03/a4/fabfdacbe252e6530f56500e95d4b092df399e2474333c13a1ae3b995f26/llama_index_readers_web-0.4.2-py3-none-any.whl", hash = "sha256:f277a5bf6f843ea095fbfbd13b1a1866cbc6ec5f998de16f3c6294a864b9f0ef", size = 101488, upload-time = "2025-06-03T17:36:15.215Z" },
 ]
 
+[[package]]
+name = "llama-index-readers-youtube-transcript"
+version = "0.3.0"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "llama-index-core" },
+    { name = "youtube-transcript-api" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/f8/f1/021d085dd5f72f0e8e2da459fe9942cf8338331691583dbc500f29416aa0/llama_index_readers_youtube_transcript-0.3.0.tar.gz", hash = "sha256:2259dc60253bef07d9a39f57665f71bc09011c9ca9774a6c5331185bd8fdff08", size = 2907, upload-time = "2024-11-18T00:00:36.334Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/ae/a9/76ba4d311212950d8acad4fb0a490d6e98e91ff7622b720b54891420f013/llama_index_readers_youtube_transcript-0.3.0-py3-none-any.whl", hash = "sha256:08aa86ed287b1f7de60bbb9e7d6e6e4ed0873b85c9a32b2af00350b900bc40c3", size = 3629, upload-time = "2024-11-18T00:00:35.316Z" },
+]
+
 [[package]]
 name = "llama-index-tools-duckduckgo"
 version = "0.3.0"
@@ -4083,6 +4098,19 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/b4/2d/2345fce04cfd4bee161bf1e7d9cdc702e3e16109021035dbb24db654a622/yarl-1.20.1-py3-none-any.whl", hash = "sha256:83b8eb083fe4683c6115795d9fc1cfaf2cbbefb19b3a1cb68f6527460f483a77", size = 46542, upload-time = "2025-06-10T00:46:07.521Z" },
 ]
 
+[[package]]
+name = "youtube-transcript-api"
+version = "1.1.0"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "defusedxml" },
+    { name = "requests" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/36/dd/10d413b20a2d14fa483853d0f6d920a0a0a6887d7c60167e4641733f99fb/youtube_transcript_api-1.1.0.tar.gz", hash = "sha256:786d9e64bd7fffee0dbc1471a61a798cebdc379b9cf8f7661d3664e831fcc1a5", size = 470144, upload-time = "2025-06-11T22:30:44.048Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/b3/69/63f1b9f96a9d3b6bd35288fe27f987c41bd157e47b3d07ca025549e3f8e6/youtube_transcript_api-1.1.0-py3-none-any.whl", hash = "sha256:876ac42b1e3f8cc99b81d8fd810bd74ed07511e51dff5db50e714e3156ad3595", size = 485739, upload-time = "2025-06-11T22:30:40.515Z" },
+]
+
 [[package]]
 name = "zipp"
 version = "3.23.0"