Skip to content
Get packs

Guide · Agent skill

How to use the Interactive Setup Wizard agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

The wizard skill lets an AI agent build a small interactive bash script that guides you through steps only a human can do, such as creating API keys in a third-party dashboard, setting up credentials, or configuring CI secrets. It is for developers and teams who would rather run a repeatable setup script than re-explain the same manual procedure to an AI each time. After this guide you will know what the skill does, how to install it, how to ask for a wizard, and what to check before you run the script it produces.

Interactive Setup Wizardmattpocock/skills · MIT

Quick start

  1. 1Download the skill. Download the wizard skill ZIP from this page.
  2. 2Unzip it into a skills folder. Unzip it into ~/.claude/skills/ for personal use, or into the project's .claude/skills/ folder.
  3. 3Or upload it in the Claude apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the skill.
  4. 4Describe the manual setup. Ask the agent in plain language for a guided wizard for the human-only steps, such as API keys or CI secrets.
  5. 5Review the stages. Check the ordered stages and the values each one produces, then confirm or change them.
  6. 6Run the script yourself. Run the generated script in your terminal and follow each prompt, confirming each stage.

What the wizard skill makes the agent do

When you ask the agent to set up something that needs a person at a browser, the skill tells it to write an interactive bash script instead of trying to do everything in one conversation. The script opens each web page, says exactly what to click and copy, and hides secret input while you type it.

The script saves values to a .env file or to GitHub secrets, asks for confirmation before each stage, and shows how many stages are left. The agent writes only the steps specific to your project, because the shared plumbing comes from a template file that the skill includes.

  • Opens each URL before asking for its value
  • Hides secret input
  • Saves values to .env or GitHub secrets (via the gh CLI)
  • Asks for confirmation before each stage and shows progress
  • Ends with a closing summary

Install the wizard skill from the ZIP or the Claude apps

Download the ZIP for the wizard skill from this page. Unzip it into your personal skills folder at ~/.claude/skills/, or into a project's .claude/skills/ folder if you want the skill to apply only to that project.

If you use the Claude apps instead, upload the skill under Settings → Capabilities → Skills.

  • Personal install: ~/.claude/skills/
  • Project install: the project's .claude/skills/ folder
  • Claude apps: Settings → Capabilities → Skills

Trigger the wizard with a natural request

You do not need a special command. Describe the manual setup in plain language and say that you want a guided script. The skill is meant for work only a human can do, so name that part of the job.

The skill's own description lists the situations it fits: provisioning infrastructure, setting up credentials or CI secrets, walking an unfamiliar third-party dashboard, and running a one-off migration or cutover. It should not be used for steps the agent can perform itself, such as editing code or running commands.

  • "Set up the .env file for this project. I need to get the API keys from the provider's dashboard, so walk me through it with a wizard."
  • "Create a wizard that helps me add our CI secrets to GitHub Actions."
  • "I'm doing a one-off database cutover next week. Build a step-by-step wizard with confirmations before each irreversible action."

Review the stages before you run the script

Before writing the script, the agent is meant to read your repo and show you the ordered list of stages and the values each one produces. You can add, drop, or reorder stages. Check that every value has a clear source, a clear destination, and a clear secret or public status.

Run the script yourself. The agent is not meant to run it end to end, because the script opens browsers and waits for your input.

  • Check where each value comes from and where it is written
  • Confirm which values are secret
  • Confirm which stages are irreversible

Safety notes: scripts, network and secrets

This skill runs scripts and uses the network. The generated script opens web pages, may write secret values to local files, and may set GitHub secrets and variables. Use it only on a machine where you are comfortable storing those values.

Avoid using it for production credentials on a shared machine. The skill is designed to ask for confirmation before irreversible actions, but you still need to read each prompt.

  • Requires bash
  • GitHub CLI (gh) is needed only for setting GitHub secrets and variables
  • shellcheck is optional and used for linting
  • You need accounts and credentials for whichever services the wizard configures

Tips for better wizards

Name the services and files involved, such as your .env.example or GitHub workflow, so the agent can find the values the project actually needs. Keep one task per stage, since each stage clears the screen and shows only the current step.

Save the script only when you want a repeatable setup path in the repo. Otherwise, delete it once the job is done. If you commit it, link it from the README so the next person runs the script.

  • Point the agent at your existing setup files
  • Keep the wizard for human-only steps
  • Commit the script only for repeatable onboarding

What people use it for

  • Setting up a project's .env file and API keys for the first time
  • Adding GitHub Actions secrets and variables with a repeatable script
  • Walking through an unfamiliar third-party dashboard step by step
  • Running a one-off migration or cutover with confirmations before each irreversible action

Frequently asked questions

How do I install the wizard agent skill?

Download the ZIP for the skill from this page and unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. In the Claude apps, you can upload it under Settings → Capabilities → Skills.

Is the wizard skill free to use?

The skill is listed under the MIT licence. Check the licence text in the source repository for the full terms.

What do I need to use the wizard skill?

You need bash. The GitHub CLI (gh) is needed only if the wizard sets GitHub secrets or variables, and shellcheck is optional for linting. You also need accounts and credentials for any third-party services the wizard configures.

Can the agent run the wizard automatically without me?

No. The generated script blocks on human input and opens browsers, so it is built for a person to run at the terminal. The skill is not meant for fully automated setup.

Does the wizard skill store my secrets safely?

The script hides secret input while you type and writes values to local files such as .env, and it may set GitHub secrets. Because it stores secret values locally, avoid using it for production credentials on a shared machine.

Why won't the agent use the wizard for editing code or running commands?

The skill says not to use it for steps the agent can perform itself. It is meant only for manual steps that require a human, such as provisioning, credentials and one-off migrations.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Use the wizard skill if you need a repeatable, checked setup script for steps that require a person at a browser, such as .env files, API keys, GitHub secrets or a one-off cutover. Pick something else if you want the agent to automate setup with no human present, or if you are handling production credentials on a shared machine.

Download the skill