If you write code with an AI assistant, deployment is often the one step you still drop back to a browser for. It does not have to be. We publish a skill for Claude Code that teaches it how to deploy to Appliku, read your logs, and manage your domains and datastores, so you can stay in your editor and just ask.

This post explains what the skill is, how to install it, and what it can and cannot do.

Table of Contents

What a skill actually is

A skill is a package of instructions that an AI coding agent loads when a task calls for it. The Appliku skill does not add magic or a hidden API. It gives Claude Code accurate, up-to-date knowledge of the appliku CLI and Python SDK: every command, the arguments each one takes, how authentication works, and the common workflows. With that loaded, the agent runs the same commands you would run by hand, instead of guessing at them or sending you to the dashboard.

In other words: the skill is knowledge, the CLI is the muscle. The agent still does the real work through the same tool and the same API token you use yourself.

Install it

The skill installs with one command:

npx skills add appliku/skill -g -a claude-code

The -g flag installs it globally so it is available in every project, and -a claude-code targets Claude Code specifically.

You also need the CLI available and a token in the environment so the agent can authenticate. The quickest path:

uv tool install appliku
export APPLIKU_TOKEN=your_api_token

Create the API token in your Appliku account settings. See the CLI post for the full install and login options.

What it can do

Once installed, Claude Code knows how to:

  • Deploy apps and stream the deployment logs back to you
  • Read application logs across one process or all of them, on server-mode and cluster-mode apps
  • Manage domains, including adding a custom domain and checking DNS propagation
  • Manage datastores and volumes, list, start, stop, restart, delete
  • Handle cron jobs, clusters, servers, SSH keys, and team invites
  • Use the Python SDK for the operations the CLI does not cover, like creating apps and managing config vars

A real example

Say a deploy just failed. Instead of opening the dashboard, you can ask in plain English:

Deploy my "storefront" app on the acme team, and if it fails, show me the last 100 lines of the web process logs.

With the skill loaded, Claude Code will list your apps to find the id, trigger the deploy, poll the deployment, and if it errors, pull the logs and summarize what went wrong, all with the CLI commands it now knows. You review the result and decide what to do next. No context-switching, no digging through the docs for the right flag.

Be deliberate about what you let it do

An agent using this skill acts with your API token, which means it can do anything you can do on that team. That is the point, and it is also the thing to be thoughtful about:

  • Read-only tasks (fetching logs, listing apps, checking DNS) are low risk. This is the best place to start.
  • Mutating actions (deploys, restarts, deleting a datastore) change real infrastructure. Keep a human in the loop for those until you trust the flow, and use a token scoped to a team where a mistake is recoverable.
  • Treat the token like a password. Do not paste it into a shared session.

None of this is unique to Appliku. It is just the reality of handing deploy access to any automated helper, and the skill does not remove your judgment from the loop.

Using a different agent?

The skill is built for Claude Code today. If you use Cursor, Windsurf, Claude Desktop, or another MCP-capable client, the same underlying CLI and SDK are available to you now, and we are building a Model Context Protocol server so those agents get first-class support too. We will link the setup guide here when it ships.

Try it

Start deploying free, install the skill, and ask your agent to ship something. If you are still comparing platforms first, the comparison hub shows where Appliku fits, and the CLI post covers everything the skill is driving under the hood.