NVIDIA AI Factory Operations Agent UI

This container provides the web interface, API, command-line interface, and Model Context Protocol (MCP) gateway for the NVIDIA AI Factory Operations Agent blueprint. It gives infrastructure operators and site reliability engineers a single interface for evidence-backed investigation of AI factory cluster health and workload failures.

Capabilities

• Conversational investigation across Kubernetes, Slurm, Base Command Manager, Prometheus, and configured operational systems.
• Access to specialized diagnostic and research agents deployed by the blueprint.
• Evidence presentation, investigation history, and downloadable diagnostic results.
• Governed MCP tool execution with explicit controls for operations that can modify infrastructure.
• Administration of supported agent connections and Research Agent document collections.

Intended use

Use this container as the user-facing service in an AI Factory Operations Agent deployment. Typical workflows include investigating unhealthy nodes or workloads, correlating infrastructure telemetry, retrieving operational documentation, and reviewing recommended remediation steps. It is an experimental operations aid and does not replace an organization's monitoring, incident-management, access-control, or change-management systems.

Deployment requirements

Deploy this image through the AI Factory Operations Agent Helm chart. The complete deployment requires a Kubernetes cluster, access to a supported OpenAI-compatible language-model endpoint, and connectivity to the operational services enabled in the chart. Individual integrations can require additional service credentials and Kubernetes RBAC configuration. The UI container itself does not require a GPU. Images are provided for Linux AMD64 and ARM64.

Installation, configuration, architecture, security controls, and support policy are documented in the AI Factory Operations Agent repository.

Publisher

Built and published by NVIDIA Corporation. Licensed under the Apache License 2.0.