Different LLMs may associate different tool names with the same intent.
A tool named search_products may be selected reliably by one model, while another may select it more reliably when it is named find_products or product_search.
WebMCP could support aliases while keeping one canonical tool identity.
document.modelContext.registerTool({
name: "search_products",
aliases: ["find_products", "product_search"],
description: "Search for products.",
inputSchema: { ... },
execute: ...
});
The exact API is just an example. The canonical name would identify the tool, while aliases could give models additional names to associate with the same tool. If an alias is invoked, it would resolve to the canonical tool.
Different LLMs may associate different tool names with the same intent.
A tool named
search_productsmay be selected reliably by one model, while another may select it more reliably when it is namedfind_productsorproduct_search.WebMCP could support aliases while keeping one canonical tool identity.
The exact API is just an example. The canonical name would identify the tool, while aliases could give models additional names to associate with the same tool. If an alias is invoked, it would resolve to the canonical tool.