Getting Started

Get AgentOps running on your Kubernetes cluster in minutes.

less than a minute

AgentOps is a Kubernetes-native platform for deploying and operating AI agents as first-class workloads. You define agents, tools, resources, and channels as Custom Resources — the operator handles deployments, jobs, storage, networking, MCP tool integration, memory, and tracing.

Where to begin

Quickstart

Install the platform, deploy your first agent, and open the console in under 5 minutes.

Installation

Full installation guide covering all configuration options, minimal installs, ingress, model providers, and uninstalling.

Architecture

Understand how the operator, console, memory service, and tracing fit together across namespaces.

What you get

The agentops-platform Helm chart deploys:

Component	Description
Operator	Kubernetes controller managing 6 CRDs (Agent, AgentRun, AgentTool, AgentResource, Channel, Provider)
Console	Go BFF + SolidJS PWA for interacting with agents, viewing traces, and managing memory
Memory	SQLite + FTS5 memory service with relevance-ranked context injection
Tempo	Distributed tracing backend — all components emit OTLP traces

Agents run in a dedicated agents namespace, isolated from platform control-plane components in agent-system. Tools are distributed as OCI artifacts and mounted into agent pods at runtime via MCP (Model Context Protocol).