⚡ OpenShift Lightspeed CLI

AI-powered OpenShift assistant from your terminal. Ask questions, troubleshoot, and manage your cluster — powered by OLS and NVIDIA NIM.

Install

npm install -g @ols-cli/lightspeed

Quick Start

# Configure NVIDIA NIM (works immediately, no OLS needed)
ols config set nimApiKey nvapi-your-key-here
ols config set nimModel meta/llama-3.1-405b-instruct

# Or configure OLS (if you have Lightspeed Service on your cluster)
ols config set serviceUrl https://lightspeed-service.openshift-operators.svc:8443

# Start chatting
ols

Usage

ols                              # Interactive REPL
ols "how do I scale?"            # One-shot query
ols health                       # Check backend health
ols conversations                # List OLS conversations
ols config show                  # Show configuration
ols config set <key> <value>     # Set config
ols help                         # Help

Backends

Backend	Setup	Description
OLS	ols config set serviceUrl <url>	OpenShift Lightspeed Service (cluster-deployed, RAG with OCP docs)
NIM	ols config set nimApiKey <key>	NVIDIA NIM — OpenAI-compatible LLM (works without OLS)

OLS is tried first, NIM is the fallback. Either one works standalone.

Config

Stored at ~/.ols/config.json:

{
  "serviceUrl": "https://lightspeed-service.openshift-operators.svc:8443",
  "nimApiKey": "nvapi-xxxxx",
  "nimModel": "meta/llama-3.1-405b-instruct",
  "nimBaseUrl": "https://integrate.api.nvidia.com"
}

Key	Default	Description
serviceUrl	""	OLS service URL
nimApiKey	""	NVIDIA NIM API key
nimModel	nvidia/llama-3.1-nemotron-70b-instruct	NIM model name
nimBaseUrl	https://integrate.api.nvidia.com	NIM API base URL

Auth

• OLS: K8S bearer token from kubeconfig (oc login)
• NIM: API key (no kubeconfig needed)

Examples

$ ols "what is OpenShift Lightspeed?"

OpenShift Lightspeed (OLS) is an AI-powered assistant that runs on
OpenShift clusters. It uses LLM backends with RAG-indexed OpenShift
documentation to answer questions about cluster operations.

via nim

$ ols "how do I scale a deployment?"

To scale a deployment in OpenShift:

  oc scale dc/myapp --replicas=3
  oc get pods -l app=myapp

For autoscaling:
  oc autoscale dc/myapp --min=2 --max=10 --cpu-percent=80

via nim

Packages

Package	npm	Description
@ols-cli/lightspeed		CLI tool
@ols-cli/client		OLS REST API + NIM client

Requirements

• Node.js >= 20.6.0
• OLS service URL or NVIDIA NIM API key
• oc CLI (for OLS auth via kubeconfig)

License

Apache-2.0