Getting Started: From Download to Working Runtime

This page exists so you can actually build the thing.

If you already work with a coding tool and a terminal, the fast track below is all you need. If terms like “repository” or “CLI” are new territory, keep scrolling — there is a guided version that explains the landscape first.

Fast Track

For people who already use Claude Code, Codex CLI, Gemini CLI, or GitHub Copilot — and who know what a terminal is.

That is it. The template includes a getting-started.md that the AI will read and walk you through. Every file has comments explaining what it does and what to change.

The Guided Path

If the fast track felt like it assumed too much, this section is for you. No shame — everyone starts somewhere, and the tools involved here are newer than most computer science degrees.

What Are We Actually Building?

A Werkbank—workbench—for your AI tools.

Right now, every time you open ChatGPT, Claude, or a coding assistant, you start from zero. The tool does not know who you are, how you work, what quality means to you, or what it should push back on. You re-explain the same things in every conversation.

The Keeper Runtime template fixes that. It is a set of text files that tell any AI tool:

• who you are and what you do
• how your partner should work (direct? careful? opinionated?)
• what quality standards apply
• what shortcuts are not acceptable

Those files get placed where each tool expects them. After that, every new conversation starts with shared context instead of a blank slate.

What You Need

A coding tool (IDE). This is the application where you open folders, read files, and talk to an AI assistant. The best starting point:

Visual Studio Code — free, runs on Mac, Windows, and Linux. Download it from code.visualstudio.com. Install it like any other application.

Once VS Code is installed, you can add an AI extension:

• GitHub Copilot (built into VS Code, requires a GitHub account)
• Continue (open source, connects to Claude or other models)
• Cline or Roo Code (autonomous coding agents inside VS Code)

Any of these will work with the template. Pick one and install it via the Extensions panel in VS Code (the icon with four squares on the left sidebar).

Setting Up Your Workspace

Before downloading anything, let’s set up a clean place to work.

What is a folder structure for code?

Developers keep all their projects in one parent folder. This is not a rule anyone enforces — it is just the pattern that works. Everything in one place, easy to find, easy to back up.

Create a folder like this:

~/repos/              ← Mac/Linux (~ means your home directory)
C:\Users\You\repos\   ← Windows

The name repos is a convention (short for “repositories”), but you can call it projects, code, or work — whatever makes sense to you.

How to create it:

• Mac: Open Finder → Go → Home → New Folder → name it repos
• Windows: Open File Explorer → navigate to C:\Users\YourName\ → New Folder → name it repos
• Or in a terminal: mkdir ~/repos (Mac/Linux) or mkdir C:\Users\You\repos (Windows)

Download and Place the Template

1. Download the template ZIP from the button above (or click here)
2. Unzip it — you will get a folder with files inside
3. Move that folder into your repos directory
4. Rename it to something meaningful, like my-runtime or keeper

Your structure should now look like this:

~/repos/
  └── my-runtime/
      ├── getting-started.md
      ├── identity/
      ├── work-mode/
      ├── skills/
      └── scripts/

Open It in VS Code

1. Open Visual Studio Code
2. File → Open Folder
3. Navigate to ~/repos/my-runtime/ and select it
4. You should see the file tree on the left

Talk to the AI

Now open the AI chat in your coding tool (the exact location depends on which extension you installed — usually an icon in the sidebar or a keyboard shortcut).

Paste this prompt:

Read getting-started.md and walk me through the setup. One file at a time.

The AI will read the template files and guide you through customizing each one — your identity, your working style, your quality standards. It will ask questions and fill in the files based on your answers.

That is the entire process. The AI does the heavy lifting. You just answer honestly.

Concepts Worth Understanding

If the terms above were unfamiliar, here is a quick reference. Not a textbook — just enough to know what is happening.

Repository (Repo)

A folder that contains a project, plus a hidden history of every change ever made. That history is managed by Git (see below). When someone says “clone the repo,” they mean: download the project folder including its full history.

For the runtime template, you do not need Git yet. The ZIP download works fine. Git becomes useful later when you want to track your changes and sync across machines.

Git

A version control system. Think of it as “infinite undo” for your files — every change is recorded, and you can go back to any previous state. Developers use it for everything.

Git is a command-line tool, but VS Code has Git support built in, so you rarely need to type Git commands directly.

IDE (Integrated Development Environment)

A fancy name for a code editor with extra features — file browser, terminal, extensions, debugging. Visual Studio Code is an IDE. So are Cursor, Windsurf, and IntelliJ. For this template, any editor that can open a folder and run an AI extension works.

CLI (Command Line Interface)

A text-based interface where you type commands instead of clicking buttons. On Mac this is Terminal, on Windows it is PowerShell or Command Prompt.

Claude Code, Codex CLI, and Gemini CLI are AI tools that run in the terminal. They are powerful but assume you are comfortable typing commands. If you are not there yet, start with VS Code and a visual AI extension.

Prompt

A message you send to an AI. It can be a question, an instruction, or a description of what you want. The quality of the prompt matters — the runtime template exists precisely because one-off prompts lose context between conversations, while a runtime gives the AI permanent context before you even start talking.

What the Template Does

The template is a set of Markdown files (.md — plain text with simple formatting). Each file covers one aspect of the working relationship:

The AI reads these files at the start of every session. That is why it stops being a blank-slate chatbot and starts being a partner with memory.

What Comes After Setup

Once the runtime is configured:

1. Use it. Open your coding tool, start a conversation. The AI now knows your context.
2. Refine it. After a few sessions, you will notice things that are missing or wrong. Edit the files directly.
3. Deploy it. When you are ready, the deploy scripts place your runtime files where each tool expects them. The Deployment deep dive covers that step.
4. Version it. Eventually, add Git to track your changes. That way you have history, can sync across machines, and never lose a good version.

The whole system is text files. No database, no cloud service, no vendor lock-in. If you can edit a text file, you can maintain your runtime.