[FEATURE] Addition of Moss as Tool or memory to AutoGen

[CoderOMaster]

Add Moss as a Tool to Autogen to enable low-latency, local-first semantic search within agents and chains.

Moss is a real-time semantic search engine optimized for conversational and multimodal AI. It enables agents, voice assistants, and chatbots to retrieve, reason, and respond in sub-10ms by colocating the vector search index with the model in the browser, on-device, or on the server. It supports hybrid keyword + semantic retrieval and comes with JS/Python SDKs.

This feature would expose Moss as a Tool that agents can call (e.g., moss_search("query")) to:

retrieve top-k results using Moss’s low-latency semantic engine,
return rich text snippets or document metadata for downstream reasoning,
optionally blend keyword and vector search with parameter control.
Available at https://usemoss.dev/