Why Groundwork
A development environment should teach judgment, not just install tools.
Groundwork exists because AI makes it easier to produce code before you understand the system around it. The answer is not to slow people down. It is to give them a setup that makes directing, inspecting, and verifying agent-assisted work feel normal from day one.
Groundwork will not make someone an agentic engineer by itself. That is a long practice of specification, review, verification, security judgment, and taste. This repo shows the arc, sets up the working environment, and gives learners their first serious reps so they start on the right path.
Start quickly, stay accountable. Groundwork gives learners and teams the shell, Git, docs, project instructions, and validation habits that keep AI-assisted development inspectable instead of mysterious.
#What makes it different
Agents are part of the workflow
The repo treats agents as collaborators you direct and supervise through files, commands, diffs, and tests.
One foundation, several environments
macOS gets the full desktop setup. Linux, WSL, and containers share the terminal layer agents can actually operate.
The docs explain the why
Tool pages earn their place by showing how they help humans direct, inspect, or verify work.
Validation is part of done
Setup scripts, rendered templates, docs links, public hygiene, and agent-adapter wiring are checked before release.
#The philosophy
Groundwork is not a terminal purity test. VS Code, Cursor, Neovim, tmux, Raycast, browsers, and future tools are choices. The durable skill is learning how to make work observable: know where files live, read the diff, run the checks, understand what changed, and keep the human as the senior partner.
The terminal matters because it is the shared operating language between humans and agents. It is fast, scriptable, inspectable, and easy to reproduce in a local shell, a WSL environment, or a headless container.