ResourceLink, EmbeddedResource Metadata Loss During MCP to LangChain Conversion

When converting MCP `ResourceLink`, `EmbeddedResource` content blocks to LangChain content blocks, important metadata fields (such as `name`, `description`, `size` and other resource properties) are being discarded. Only `uri` and `mimeType` are preserved, resulting in information loss. ## Problem Description In `tools.py`, the `_convert_mcp_content_to_lc_block()` function handles `ResourceLink` conversion as follows: ```python if isinstance(content, ResourceLink): mime_type = content.mimeType or None if mime_type and mime_type.startswith("image/"): return create_image_block(url=str(content.uri), mime_type=mime_type) return create_file_block(url=str(content.uri), mime_type=mime_type) ``` ### Current Behavior - Only `uri` and `mimeType` from `ResourceLink` are converted - Resource name, description and other metadata fields are **completely discarded** - The resulting LangChain content blocks lack this contextual information ### Expected Behavior - All available metadata from `ResourceLink` should be preserved during conversion - LangChain content blocks should include: - `name` field (if available in ResourceLink) - Any other relevant resource metadata - Full fidelity representation of the original MCP content ## Impact - **Information Loss**: Downstream tools and models lose context about resource names and properties - **Poor User Experience**: File/image blocks lack descriptive information - **Reduced Functionality**: MCP servers can't effectively communicate resource metadata to LangChain agents ## Affected Code Location - **File**: `langchain_mcp_adapters/tools.py` - **Function**: `_convert_mcp_content_to_lc_block()` - **Lines**: ResourceLink handling block ## Steps to Reproduce 1. Create an MCP tool that returns `ResourceLink` content with a `name` field 2. Convert the tool to LangChain format using `convert_mcp_tool_to_langchain_tool()` 3. Call the tool and inspect the returned content blocks 4. Observe that the `name` field …