Skip to content

Editions

A Starkite edition is a packaged binary containing a specific set of runtime modules. Providing multiple editions allows you to minimize binary size and reduce dependency graphs in space-constrained or security-sensitive environments. All editions share a common language core and base modules.

By default, the all-in-one kite binary is used. Reach for a lean edition only if binary footprint or dependency auditing constraints require it.

Binary Included Modules Use Case
kite Base + Kubernetes + MCP Default all-in-one developer workstation binary
kitecmd Base only Minimal system scripting, general automation, and CI tasks
kitecloud Base + Kubernetes (k8s module + kite kube CLI commands) Cloud-native ops and manifest workflows
kiteai Base + MCP capabilities Agentic AI tool integration and MCP server/client

Lean editions (kitecmd, kitecloud, and kiteai) are strict subsets of kite, packaged smaller for space-conscious environments like init containers, edge nodes, and CI runners.

Per-edition Go modules

To optimize binary footprints, each edition is defined as a distinct Go module with independent dependency graphs. This ensures omitted modules are never compiled in. For example, kitecmd links no Kubernetes or MCP libraries, producing a lean binary with no cloud-native library overhead. Lean editions are tailored for container init steps, edge computing, or CI runners; developers typically use the all-in-one kite binary.

Base modules

All editions include the following 28 base modules:

os, fs, fmt, io, vars, strings, regexp, json, yaml, csv, sql, time, base64, hash, uuid, template, gzip, zip, log, concur, retry, table, runtime, ssh, http, fleet, inventory, test

For module details, see the API Reference.

What kitecloud adds

The kitecloud edition adds the k8s module and the kite kube CLI command set. This enables Kubernetes CRUD operations, controller runtimes, custom resource definitions (CRDs), and admission webhooks.

For guides, see the Kubernetes Guide.

What kiteai adds

The kiteai edition adds Model Context Protocol (MCP) support (mcp module). This enables scripts to act as MCP tool servers (mcp.serve()) or connect to remote tool servers as clients (mcp.connect()) over stdio and HTTP.

For guides, see the MCP Server Guide.