Skip to main content

​
What You’ll Do

Open the Meta Agent, explore your infrastructure, and execute your first task through natural language. Time: ~5 minutes

​
Before You Begin

​
Step 1: Open the Meta Agent

The Meta Agent is your primary interface for exploring and interacting with Kubiya. Two ways to access:
  1. Compose Workspace - Click Compose in the sidebar for the full-screen workspace at /chat
  2. Quick Panel - Press Cmd+J (Mac) or Ctrl+J (Windows/Linux) from any page
Compose Workspace The Compose workspace provides a dedicated space for conversations with conversation history in the sidebar. Use the Quick Panel for fast queries without leaving your current page.
For detailed UI navigation, see the Web Interface Guide.
On first use, select the categories most relevant to your role (Security, Infrastructure, Cost Optimization, etc.).

​
Step 2: Try Your First Query

Ask the Meta Agent a simple question: 
"What environments are available?"

"List all connected data sources"

"Show me the status of my task queues"
The Meta Agent has access to all Kubiya platform capabilities through MCP tools.

​
Step 3: Set Up a Connector

Connectors enable agents to act on external systems. Go to Infrastructure → Connectors and add your first connection:
ConnectorDescription
AWSCloud resource management
GitHubRepository access, CI/CD
SlackTeam messaging
JiraIssue tracking
KubernetesContainer orchestration
Without connectors, agents can explore and analyze but cannot take actions on external systems.

​
Step 4: Add an Ingestion Source

Ingestion Sources populate the Context Graph with data about your infrastructure. Go to Context Graph → Data Ingestion and add your first source:
  • Cloud Providers: AWS, Azure, GCP
  • Identity: Okta, Microsoft Entra ID
  • DevOps: GitHub, Kubernetes, CircleCI
  • Custom: Upload CSV/JSON files
The Context Graph starts empty. Add ingestion sources so the Meta Agent can explore your infrastructure.

​
Step 5: Execute a Task

Back in the Meta Agent, try more advanced queries: 
"What EC2 instances are running in production?"

"Show me all IAM roles with admin access"

"What services depend on the auth database?"
Use the Remote Task Queues button in the chat to route execution to specific workers.

​
What You Used

ComponentPurpose
Meta AgentUnified interface for exploration and orchestration
ConnectorsCredentials for agents to act on external systems
Ingestion SourcesData that populates the Context Graph
Hosted Control PlaneZero-setup coordination layer
Hosted WorkersIsolated execution for your tasks

​
Next Steps

Meta Agent

Deep dive into natural language orchestration

Connectors

Enable agents to act on external systems

Data Ingestion

Populate your Context Graph with infrastructure data

Task Queues

Distributed execution with remote workers

​
Advanced: Specialized Agents & Teams

When you need domain-specific automation beyond the Meta Agent:
  • Agents - Create specialized AI agents for specific domains (DevOps, Security, FinOps)
  • Teams - Coordinate multiple agents for complex multi-domain operations
Most users interact through the Meta Agent for general orchestration. Create custom agents only when you need specialized, reusable automation.