Command Palette

Introduction

As a DevOps engineer or SRE, you perform dozens of operational tasks every day, including checking pod status, analyzing metrics, reviewing deployments, inspecting logs, and managing configurations. Each task involves context-switching between tools, remembering specific commands, and manually piecing together information from different sources.

Ops Mode changes this.

Ops Mode is Voyager's AI-powered operations assistant designed for your day-to-day infrastructure work. Instead of jumping between kubectl, Grafana, Prometheus, AWS console, and other tools, you simply describe what you need in natural language, and Voyager executes the task across your entire stack.

Combined with the Command Palette, a library of 304 pre-built operational prompts—Ops Mode gives you instant access to common DevOps workflows without writing commands or remembering syntax.

Ops Mode at a Glance

---

304 pre-built operational prompts	Ready-to-use commands for common workflows
12 platform integrations	Kubernetes, AWS, Prometheus, Grafana, ArgoCD, and more
10 workflow categories	Daily Operations, Analytics, Incident Response, and more

---

What is Ops Mode?

Ops Mode is your AI-powered command center for infrastructure operations. It brings together all your tools—Kubernetes, Prometheus, Grafana, AWS, ArgoCD, and more—into a single natural language interface.

The Idea Behind Ops Mode

Imagine having a senior SRE sitting next to you who knows every kubectl command, every PromQL query, every AWS CLI flag, and can execute them instantly across your entire stack. That is what Ops Mode provides.

Instead of:

• Switching between terminal, browser, and console

• Remembering syntax for each tool

• Manually correlating data from different sources

• Copy-pasting outputs between systems

You simply describe what you need, and Ops Mode handles the execution.

How Ops Mode Works

Every request follows a simple flow:

1

You Describe Your Task

Express what you need in plain English:

• "Show me pods that restarted in the last hour"

• "What's the P99 latency for the checkout service?"

• "List all failed GitHub Actions in the last day"

You don't need to know the specific commands or tools involved.

2

Voyager Classifies and Routes

Based on your query, Voyager determines:

• Query type: Is this a fact lookup, a list request, or an exploratory analysis?

• Relevant platforms: Which tools are needed (Kubernetes, Prometheus, GitHub, etc.)?

• Execution plan: What sequence of operations will get you the answer.

3

Agents Execute

Specialized agents handle the actual work:

• The Kubernetes agent runs kubectl operations

• The Prometheus agent executes PromQL queries

• The AWS agent queries CloudWatch and other services

Each agent knows its domain deeply and executes efficiently.

4

Results Are Formatted

Voyager returns results in a clean, actionable format:

• Tables for list data (pods, deployments, alerts)

• Summaries for analytical queries

• Direct answers for fact-based questions

• Evidence citations showing where data came from

---

Use Cases

Here are real-world scenarios showing how DevOps engineers and SREs use Ops Mode and the Command Palette.

Scenario 1: Morning Standup Preparation

The situation: It's 8:45 AM. You have standup in 15 minutes and need to know the current state of production—any overnight issues, current deployments, system health.

With Ops Mode: Ask a single query (e.g., "Give me a production health summary...") and get a concise summary showing pods, alerts, deployments, and error rates within seconds.

Scenario 2: Debugging Slow API Response

The situation: A developer reports that the /api/orders endpoint is slow.

With Ops Mode: Ask an analysis query combining Prometheus, Kubernetes, and deployment history. You receive correlated evidence (latency spike, high pod CPU, recent deployment), surfacing the likely cause.

Scenario 3: Pre-Deployment Checklist

The situation: Preparing to deploy the payment service.

With Ops Mode: Use a pre-deployment prompt to verify pod health, replicas, resource usage, Sentry error rates, and active alerts. Get a deployment-ready summary quickly.

Scenario 4: Investigating OOMKilled Pods

The situation: Pods are getting OOMKilled.

With Ops Mode: Query for OOMKilled pods in a timeframe and receive which pods, how often, memory limits, and usage before kill—plus actionable recommendations.

Scenario 5: Capacity Planning Review

The situation: End-of-quarter resource utilization reporting.

With Ops Mode: Run analytics prompts to gather trends, comparisons, and hotspot lists for presentations and planning.

---

Frequently Asked Questions

Can I perform write operations (changes) in Ops Mode?

Yes, but with safeguards. Read operations execute immediately. Write operations (scaling, patching, syncing, deployments) require explicit confirmation before execution. You're always in control.

How do I know which platform agents are being used?

Voyager automatically routes your query to the appropriate agents based on what you're asking. You can also explicitly select agents if you want to limit the scope.

Is my data safe?

Ops Mode uses secure, authenticated connections to your infrastructure. It only accesses systems you've authorized and integrated. All operations are logged for audit purposes.

Can multiple team members use Ops Mode simultaneously?

Yes. Ops Mode supports multi-user access. Each user's queries are processed independently, and all operations are logged with user attribution for audit purposes.