MCP Apps are an extension direction for the Model Context Protocol (MCP) that lets servers expose user-facing application experiences, not only tools and resources. The goal is to support richer interactions between users, AI applications, and MCP-connected systems.
Traditional MCP servers expose:
- tools that an AI application can call;
- resources that provide data for context; and
- prompts that package reusable workflows.
MCP Apps extend that model toward server-rendered or server-defined user interfaces. This is useful when a tool call needs more than a hidden API operation, such as reviewing a complex object, confirming a configuration, filling a structured form, or interacting with a task-specific UI.
MCP Apps are especially relevant for enterprise AI agents because not every workflow should be handled through free-form chat. A server-controlled interface can make permissions, data provenance, and user choices clearer.
They are also related to MCP Elicitation, but the scope is different. Elicitation asks for specific user input during a protocol operation. MCP Apps can define broader interactive experiences.
Security remains central. A client should clearly show which MCP server is presenting an interface, what data is being requested or changed, and what permissions apply. Servers must not use UI controls to bypass user consent or application policy.
MCP Apps are part of the broader MCP Extensions direction, allowing richer capabilities to evolve without bloating the core protocol.
The MCP maintainers identified MCP Apps as a first-class extension area in the 2026-07-28 release candidate.
The LLM Knowledge Base is a collection of bite-sized explanations for commonly used terms and abbreviations related to Large Language Models and Generative AI.
It's an educational resource that helps you stay up-to-date with the latest developments in AI research and its applications.