[FEAT]: Support MCP Apps (SEP-1865) — Interactive UI rendering from MCP servers #3205

Open
opened 2026-02-28 06:33:26 -05:00 by deekerman · 0 comments
Owner

Originally created by @MarekFabinski on GitHub (Feb 24, 2026).

Summary

AnythingLLM already supports MCP servers as a tool client — this request is to
add support for MCP Apps (SEP-1865), the extension that allows
MCP servers to deliver sandboxed interactive UIs (HTML/JS iframes) directly
inside the chat interface.

What is it?

SEP-1865 is an official MCP extension co-authored by Anthropic and OpenAI
maintainers. It lets MCP tool servers return a
ui:// resource (rendered in a sandboxed iframe) instead of plain text,
with bidirectional JSON-RPC communication between the UI and the host.

Use cases this unlocks

  • Inline forms, approval flows, and data pickers returned by agent tools
  • Live dashboards or results grids rendered inside the conversation
  • Rich outputs from connected enterprise systems (CRM, ERP, etc.)

Request

  • Detect ui:// resource payloads in MCP tool responses
  • Render them in a sandboxed iframe within the chat message area
  • Pass user interactions back to the agent as tool output
  • Works in both Desktop and Docker deployments
Originally created by @MarekFabinski on GitHub (Feb 24, 2026). ## Summary AnythingLLM already supports MCP servers as a tool client — this request is to add support for **MCP Apps (SEP-1865)**, the extension that allows MCP servers to deliver sandboxed interactive UIs (HTML/JS iframes) directly inside the chat interface. ## What is it? SEP-1865 is an official MCP extension co-authored by Anthropic and OpenAI maintainers. It lets MCP tool servers return a `ui://` resource (rendered in a sandboxed iframe) instead of plain text, with bidirectional JSON-RPC communication between the UI and the host. - Spec: https://modelcontextprotocol.io/community/seps/1865-mcp-apps-interactive-user-interfaces-for-mcp - Official SDK & repo: https://github.com/modelcontextprotocol/ext-apps ## Use cases this unlocks - Inline forms, approval flows, and data pickers returned by agent tools - Live dashboards or results grids rendered inside the conversation - Rich outputs from connected enterprise systems (CRM, ERP, etc.) ## Request - Detect `ui://` resource payloads in MCP tool responses - Render them in a sandboxed iframe within the chat message area - Pass user interactions back to the agent as tool output - Works in both Desktop and Docker deployments
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
starred/anything-llm#3205
No description provided.