Exploring Tool Use in Large Language Models through Model Context Protocol (MCP)
Bachelor & Master Thesis
Description:
The Model Context Protocol (MCP) is a new standard for structuring prompts and interactions with large language models. This thesis explores how MCP can improve usability, control, and reliability in LLM-based applications. Students will analyze how developers describe LLM tasks and tool functions within MCP, and propose refined design patterns to make these descriptions more effective and consistent.
Requirements:
Python and LLM API experience (e.g., OpenAI, Claude, Mistral)
Familiarity with prompt engineering or conversational agents