Excalibur of the AI Era: MCP Client Software Illustrated by `Cherry studio`
Speaker
Time
Friday, January 23, 2026 4:30 PM - 5:30 PM
Venue
A3-3-301
Online
Zoom 815 762 8413
(BIMSA)
Abstract
As large language models transition from passive conversational systems to active, tool-augmented agents, the role of client software becomes increasingly critical. The Model Context Protocol (MCP) introduces a standardized framework for connecting models with external tools, resources, and execution environments. Within this framework, **MCP clients** emerge as the practical interface through which users, models, and systems interact.
This report focuses on MCP client software (called **MCP host**), using **chat-box–style applications such as Cherry Studio** as primary examples, while also examining related systems including **LobeHub/LM Studio/Anything LLM**, and other MCP-capable or MCP-adjacent tools. Beyond chat interfaces, the report surveys MCP integration and client-like roles in **programming environments (Cursor, Zed)**, **knowledge and note-taking systems (Obsidian, OpenWork)**, **desktop automation tools (Raycast)**, and **next-generation terminals (Warp)**.
By analyzing these tools from a unified perspective, the report argues that MCP clients function as the “saber” of the AI era: a controlled yet powerful extension that transforms raw model intelligence into effective, safe, and user-directed action. The discussion highlights architectural patterns, interaction models, and emerging design principles that define MCP clients as a foundational layer in modern AI systems.
This report focuses on MCP client software (called **MCP host**), using **chat-box–style applications such as Cherry Studio** as primary examples, while also examining related systems including **LobeHub/LM Studio/Anything LLM**, and other MCP-capable or MCP-adjacent tools. Beyond chat interfaces, the report surveys MCP integration and client-like roles in **programming environments (Cursor, Zed)**, **knowledge and note-taking systems (Obsidian, OpenWork)**, **desktop automation tools (Raycast)**, and **next-generation terminals (Warp)**.
By analyzing these tools from a unified perspective, the report argues that MCP clients function as the “saber” of the AI era: a controlled yet powerful extension that transforms raw model intelligence into effective, safe, and user-directed action. The discussion highlights architectural patterns, interaction models, and emerging design principles that define MCP clients as a foundational layer in modern AI systems.
Speaker Intro
Congwei Song received the master degree in applied mathematics from the Institute of Science in Zhejiang University of Technology, and the Ph.D. degree in basic mathematics from the Department of Mathematics, Zhejiang University, worked in Zhijiang College of Zhejiang University of Technology as an assistant from 2014 to 2021, from 2021 on, worked in BIMSA as asistant researcher. His research interests include machine learning, as well as wavelet analysis and harmonic analysis.