Skip to main content

Overview

The gsd:new-project command initializes a new project through a unified flow: questioning → research (optional) → requirements → roadmap. This is the starting point for any new Get Shit Done project.

Usage

/gsd:new-project
/gsd:new-project --auto

Parameters

--auto
flag
Automatic mode. After config questions, runs research → requirements → roadmap without further interaction. Expects idea document via @ reference.

How It Works

The command executes a unified workflow that:
  1. Questions - Gathers initial project context through adaptive questioning
  2. Research (optional) - Conducts domain research if needed
  3. Requirements - Documents scoped requirements based on your inputs
  4. Roadmap - Creates a phase structure for execution
All workflow gates (validation, approvals, commits, routing) are preserved during execution.

Files Created

This command creates a complete .planning/ directory structure:
.planning/PROJECT.md
file
Project context document - the source of truth for what you’re building
.planning/config.json
file
Workflow preferences and configuration settings
.planning/research/
directory
Domain research directory (created when research is needed)
.planning/REQUIREMENTS.md
file
Scoped requirements document defining what needs to be built
.planning/ROADMAP.md
file
Phase structure defining the execution plan
.planning/STATE.md
file
Project memory tracking progress and state

Examples

Interactive Mode

/gsd:new-project
Runs the full initialization flow with interactive questioning at each stage.

Automatic Mode

/gsd:new-project --auto @project-idea.md
Runs the entire flow automatically after initial config questions, using the referenced idea document as context.
In automatic mode, you must provide an idea document via @ reference for the system to work with.

Next Steps

After running gsd:new-project, start working on your first phase:
/gsd:plan-phase 1
This will create a detailed execution plan for phase 1 of your roadmap.

Allowed Tools

This command has access to:
  • Read - Read existing files
  • Bash - Execute shell commands
  • Write - Create new files
  • Task - Delegate to specialized agents
  • AskUserQuestion - Interactive questioning