Goose AI agent and MCPUI demo: GitHub issues and interactive UIs
A GitHub Universe 2025 demo shows the Goose AI agent using the GitHub MCP server to create and assign GitHub issues from the terminal, then rendering a fully interactive team project calendar via MCPUI.
Score breakdown
Developers and engineering managers can use Goose with the GitHub MCP server and MCPUI today to automate issue management and surface team workload data through interactive visual interfaces — going beyond text-only agent responses.
- 01Demo was presented at GitHub Universe 2025
- 02Goose AI agent connects to the GitHub MCP server to perform actions from the terminal
- 03The demo creates a GitHub issue titled 'speak at GitHub Universe' and assigns it to a user in the 'square up developer programs' repository
Presented at GitHub Universe 2025, this demo walks through two capabilities of the Goose AI agent integrated with the GitHub MCP server. In the first part, a natural-language prompt is issued from the terminal to create a GitHub issue titled "speak at GitHub Universe," assign it to a specific user ("Black Girl Bites"), and place it in the "square up developer programs" repository. Goose invokes the GitHub MCP server's `issue_write` tool call, completes the action, and returns a URL that confirms the issue was successfully created.
After enabling the appropriate extensions, the presenter asks Goose to display the team's project calendar.
The second part of the demo introduces MCPUI, described as an interface layer that enables AI agents to respond with fully interactive visual interfaces instead of plain text. After enabling the appropriate extensions, the presenter asks Goose to display the team's project calendar. Rather than returning a text-based list, Goose renders an interactive calendar UI. The demo highlights an "analyze workload" button within that calendar, which, when clicked, evaluates ticket assignments and reports which team members have lighter or heavier workloads — illustrating how MCPUI can surface actionable, structured information through a visual agent response.
Key facts
- 01Demo was presented at GitHub Universe 2025
- 02Goose AI agent connects to the GitHub MCP server to perform actions from the terminal
- 03The demo creates a GitHub issue titled 'speak at GitHub Universe' and assigns it to a user in the 'square up developer programs' repository
- 04Goose uses a tool call called `issue_write` to create the issue and returns a confirmation URL
- 05MCPUI is an interface layer that lets AI agents respond with interactive visual UIs instead of plain text
- 06When prompted, Goose renders a fully interactive team project calendar
- 07The calendar includes an 'analyze workload' feature that identifies team members with light or heavy workloads based on assigned tickets