Update `vllm_ask_image` to `vllm_ask_image_tool`, add `Context` support, and implement asynchronous image query handling in tools.

src/gaia_solving_agent/tools.py

@@ -8,6 +8,7 @@ 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 workflows import Context
 
 from gaia_solving_agent import TAVILY_API_KEY, NEBIUS_API_KEY, MISTRAL_API_KEY
 
@@ -26,6 +27,27 @@ async def tavily_search_web(query: str) -> str:
     client = AsyncTavilyClient(api_key=TAVILY_API_KEY)
     return str(await client.search(query))
 
+async def vllm_ask_image_tool(ctx: Context, query: str) -> str:
+    """
+    Asynchronously processes a visual-linguistic query paired with image data
+    and returns corresponding results. This function leverages visual
+    understanding and language processing to answer the provided query based
+    on the content of the given image(s).
+
+    Parameters:
+    ctx: Context
+        LlamaIndex Workflow Context to search images into
+    query: str
+        The question or request related to the provided image(s).
+
+    Returns:
+    str
+        The result or response to the provided query based on the processed
+        image content.
+    """
+    images = await ctx.get("additional_file")
+    return await vllm_ask_image(query=query, images=images)
+
 
 async def vllm_ask_image(query: str, images: ImageDocument | list[ImageDocument]) -> str:
     """