What do I need to get started with MCP Apps?

You need an OpenSearch UI application with an Observability workspace, a compatible agentic IDE (like Claude Desktop or VS Code), Node.js 22 or later, and AWS credentials with es:ESHttpGet and es:ESHttpPost permissions.

How do MCP Apps ensure the visualizations are trustworthy?

Visualizations are generated server-side by executing code against your actual data sources. This means results are deterministic and match what you would see in your OpenSearch dashboards, not just the AI's interpretation.

Large Language ModelsAmazon AI BlogAug 25, 2026Aug 26, 2026, 04:01 JST

Amazon OpenSearch MCP Apps bring interactive dashboards into AI agent chats

Send on LINE

3 Key Points

What happened
Amazon OpenSearch Service now supports MCP Apps, which extend the Model Context Protocol so each tool call returns both a text summary and an interactive visualization (trace waterfall, service map, log pattern view) rendered in your AI assistant's chat window.
Why it matters
The verification gap—reading an agent's root cause hypothesis and manually checking it in separate browser tabs—has been the bottleneck. MCP Apps let engineers verify inline in the same thread, without leaving the IDE or re-running queries.
What to watch
Setup requires a compatible agentic IDE (Claude Desktop, VS Code GitHub Copilot, Goose, ChatGPT, or Cursor), Node.js 22 or later, and AWS credentials with es:ESHttpGet and es:ESHttpPost permissions. The server runs locally, keeping data in your AWS account.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Get news like this every morning, free

Context & Analysis

This launch addresses a core friction point in agentic observability: the time lost when engineers must switch from an AI assistant's text answer to separate dashboards for manual verification. By embedding interactive visualizations directly into the conversation thread, MCP Apps aim to keep the human in the loop without breaking flow. The local MCP server preserves the control and cost-efficiency that organizations running agentic observability locally have chosen, while adding the ease-of-use typically found in vendor-hosted solutions. The dual response pattern—structured text for the agent's reasoning, plus a rendered widget for the human—bridges the gap between automated analysis and human judgment. This could make agentic workflows more practical for on-call engineers, who can now investigate, verify, and even trigger remediation without ever leaving the IDE.

FAQ

What do I need to get started with MCP Apps?: You need an OpenSearch UI application with an Observability workspace, a compatible agentic IDE (like Claude Desktop or VS Code), Node.js 22 or later, and AWS credentials with es:ESHttpGet and es:ESHttpPost permissions.
How do MCP Apps ensure the visualizations are trustworthy?: Visualizations are generated server-side by executing code against your actual data sources. This means results are deterministic and match what you would see in your OpenSearch dashboards, not just the AI's interpretation.

Amazon AI BlogRead Original Article

Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.