MCP Apps: Extending the Frontier — Ido Salomon & Liad Yosef
MCP Apps let servers return sandboxed, interactive UI instead of text alone while the host retains control of prompts and tool calls—a useful pattern for agent workflows needing charts or controls.
MCP Apps lets an MCP tool link to an HTML resource that a compatible host renders as a **sandboxed interactive view**. UI events travel back to the host, which decides whether to issue a prompt, tool call, resource request, or other action.
For agent-facing tools, return a focused chart or control when text obscures the task. Keep mutations behind host-mediated tool calls, and consider the official SDK so spec changes flow into the implementation. The same app code is presented as portable across supporting hosts.
MCP Apps lets an MCP tool link to an HTML resource that a compatible host renders as a **sandboxed interactive view**. UI events travel back to the host, which decides whether to issue a prompt, tool call, resource request, or other action. For agent-facing tools, return a focused chart or control when text obscures the task. Keep mutations behind host-mediated tool calls, and consider the official SDK so spec changes flow into the implementation. The same app code is presented as portable across supporting hosts. The specification is **still evolving**: reusable view tools and interoperability with declarative or generative UI approaches remain active work. Portability also depends on each target host implementing MCP Apps and its interaction protocol consistently.
This extends MCP from tool invocation into host-governed interactive interfaces, making charts and controls first-class results without giving embedded views direct authority over mutations. Against the candidates, that separation becomes especially important for deployment and purchasing, while the evolving specification and uneven host support narrow claims of portability and favor SDK-managed implementations.