Amazon OpenSearch Service Introduces MCP Apps for Agentic Observability

Amazon OpenSearch Service has introduced MCP Apps, a feature designed to improve observability workflows by extending the Model Context Protocol (MCP). Previously, engineers using AI agents for troubleshooting often faced a verification gap, requiring them to switch from their IDE to a separate browser-based observability tool to confirm findings.

With MCP Apps, AI agents can now return a dual response: a text-based summary and an interactive visualization, such as a trace waterfall, service map, or log pattern view. These visualizations are rendered directly within the AI assistant’s chat window. Because the visualizations are generated by executing queries against the same data sources used by OpenSearch dashboards, the results are deterministic. This allows engineers to verify agent findings without leaving their development environment. The architecture utilizes a local MCP server to bridge the IDE and the OpenSearch UI application, ensuring that data remains within the user's AWS account.