Quickstart

This guide gets gald3r installed in a project and walks you through your first commands.

Prerequisites

• A git repository (gald3r works best when your project is under version control - the .gald3r/

control plane and your commits form the audit trail).

• An AI coding assistant that gald3r supports (Claude Code, Cursor, GitHub Copilot, Codex, Gemini

CLI, and ~20 others - see the platform skills for the full list).

• Windows users: PowerShell. gald3r's helper scripts and hooks are PowerShell-based.

Step 1 - Install gald3r in your project

Open your project in your AI assistant and run the setup command:

@g-setup

This initializes the slim .gald3r/ layout in your project root:

.gald3r/
  .identity            # project id, name, type, gald3r version
  TASKS.md             # active task index
  PLAN.md              # master plan
  PROJECT.md           # mission and goals
  CONSTRAINTS.md       # active project constraints
  BUGS.md              # bug index
  SUBSYSTEMS.md        # subsystem registry
  IDEA_BOARD.md        # captured ideas
  FEATURES.md          # feature index
  tasks/               # individual task files
  bugs/  features/  subsystems/  reports/  logs/

@g-setup also runs a git readiness check - if the directory is not yet a git repo, it offers to initialize one (gald3r autopilot features expect git to be present).

Step 2 - Check status

@g-status

Shows your session context: mission, active tasks, goals, constraints, and ideas. This is also what the session-start hook surfaces automatically each time you open the project.

Step 3 - Create your first task

@g-task-add "Add a health-check endpoint to the API"

This creates a task file with a sequential ID (e.g. T001), acceptance criteria, and a status of pending, and adds a row to TASKS.md. Describe the work; the task manager handles the bookkeeping.

Step 4 - Implement and review

Run the implementation pipeline:

@g-go-code

This drives implementation against your open task specs. To also auto-review the work with a fresh independent reviewer agent (adversarial QA), use the full pipeline instead:

@g-go

Phase 1 implements; Phase 2 spawns a separate reviewer that checks the work against the acceptance criteria with no access to the implementer's reasoning.

Step 5 - Commit

@g-git-commit

Produces a structured commit following gald3r conventions (type prefix, task reference, clean trailers). gald3r commits freely but never pushes without your confirmation.

Where to go next

• Commands - the full command catalog, grouped by category.
• Agents - the specialized personas you can invoke with @g-agnt-*.
• Skills and Rules - the capability modules and conventions that power

the commands.

Detailed per-command argument and syntax reference is being expanded in a follow-up pass. For now, each index page links to the canonical source file on GitHub.

---

Last updated: 2026-06-20 (gald3r v2.1.0)