AI Agents & Agentic Experience (AX)

Kubeletto is designed with an Agentic-first interface. Modern AI coders (like Cursor, Claude, and Gemini) can easily check status, inspect logs, and deploy applications using the Kubeletto CLI or API.

1. Initializing Agent Workspace Skills

Initialize agent guides and instructions directly in your repository directory so your local AI model knows how to safely manage Kubeletto resources:

Available targets include:

• generic: Generates .agents/skills/kubeletto/SKILL.md and .agents/AGENTS.kubeletto.md.
• cursor: Generates .cursorrules and a detailed kubeletto.mdc rules file.
• claude: Generates CLAUDE.md command references.

2. Safety and Non-Interactive Usage

AI agents run scripts in headless environments where standard confirmation prompts would hang. To prevent this, Kubeletto CLI enforces:

• Automatic Prompts Guard: If standard input is not a TTY (character device) or the --non-interactive flag is present, any command that asks for confirmation (like rollback or keys revoke) fails immediately with exit code 2 unless --yes / -y is provided.
• JSON Output: Commands support --output json which routes progress text logs to stderr and leaves stdout clean for raw JSON serialization.