[FEATURE] Helm Chart: Add support for enabling metrics endpoint and optional ServiceMonitor #1369

📋 Prerequisites

• I have searched the existing issues to avoid creating a duplicate
• By submitting this issue, you agree to follow our Code of Conduct

📝 Feature Summary

Add Helm values to easily enable the controller metrics endpoint and optionally deploy a Prometheus Operator ServiceMonitor.

❓ Problem Statement / Motivation

The kagent controller already supports exposing a metrics endpoint via the flags:

• --metrics-bind-address
• --metrics-secure

These can also be configured via environment variables, for example:

export METRICS_BIND_ADDRESS=":9093"
export METRICS_SECURE="false"

However, the current Helm chart does not provide a simple, documented way to enable and expose this metrics endpoint.

Even if we manually inject environment variables through values like:

controller:
  env:
    - name: "METRICS_BIND_ADDRESS"
      value: ":9093"
    - name: "METRICS_SECURE"
      value: "false"

This is still incomplete because:

1. The Helm chart does not automatically expose the metrics port in the controller Service.
2. Users must manually create or modify a Kubernetes Service to reference the new port.
3. Users must manually create a ServiceMonitor (when using Prometheus Operator) to scrape the endpoint.
4. There is no single, cohesive way to “enable metrics” through values.yaml.

This makes observability setup more complex than necessary.

💡 Proposed Solution

Introduce Helm values for metrics configuration and optional Prometheus Operator integration.

For example:

metrics:
  enabled: true
  bindAddress: ":9093"
  secure: false

  service:
    enabled: true
    port: 9093
    type: ClusterIP

  serviceMonitor:
    enabled: true
    interval: 60s
    scrapeTimeout: 10s
    labels: {}

🔄 Alternatives Considered

No response

🎯 Affected Service(s)

Controller Service

📚 Additional Context

Many Kubernetes projects provide a built-in metrics and serviceMonitor configuration block in their Helm charts. Adding similar support to kagent would:

1. Improve user experience
2. Reduce configuration drift
3. Encourage consistent observability setups
4. Make kagent easier to integrate into Prometheus-based monitoring stacks

🙋 Are you willing to contribute?

• I am willing to submit a PR for this feature