🎉 gald3r v5.0.0-beta.51 is out — Throne, gald3r_ide and the CLI ship together on one tag — Throne sign-in lands on the Wails build, and every project now carries its first commandment.See release notes →

Wiki documentation — Gald3r-Labs/gald3r

gald3r Wiki

Last synced July 6, 2026Edit on GitHub wiki

gald3r Wiki

Welcome to the gald3r wiki - the user guide for the gald3r framework.

What is gald3r?

gald3r is an AI-agent project operating system. You install it into any code project, and it gives your AI coding assistant (Claude Code, Cursor, Copilot, Codex, and 20+ other platforms) a shared, file-based control plane for tasks, plans, bugs, features, constraints, and cross-project coordination - all stored as plain markdown under a .gald3r/ folder in your repo.

The goal: turn ad-hoc "vibe coding" into a repeatable, auditable workflow. Every piece of work is a task with acceptance criteria. Every change can be reviewed by a fresh independent agent. Every decision leaves a trail in git.

Key Concepts

gald3r is built from six kinds of building blocks. Each has its own index page in this wiki:

ConceptWhat it isIndex page
TasksThe atomic unit of work. Each task is a markdown file with acceptance criteria, status, and history. Managed via @g-task-* commands.(see Commands -> Task Management)
Commands@g-* slash commands you invoke to do things - create a task, run a code review, plan a release, check status.Commands
SkillsReusable capability modules that commands and agents load on demand (e.g. git commits, bug tracking, web crawling).Skills
AgentsSpecialized personas (@g-agnt-*) that own a slice of the workflow - task manager, code reviewer, verifier, planner.Agents
HooksLifecycle scripts that fire automatically (session start, agent complete, pre-commit) to inject context and enforce gates.Hooks
RulesAlways-on guidance loaded every session - git workflow, documentation standards, enforcement gates.Rules

How a typical session works

  1. Session start - a hook injects your project's .gald3r/ context (mission, active tasks,

constraints) so the agent knows where things stand.

  1. Create work - @g-task-add "Add OAuth login" files a task with acceptance criteria.
  2. Implement - @g-go-code (or @g-go) drives implementation against the task spec.
  3. Review - a fresh independent reviewer agent verifies the work against the acceptance

criteria (PASS / FAIL), with no access to the implementer's reasoning.

  1. Commit - @g-git-commit produces a structured, convention-following commit.
  2. Track - @g-status and @g-report show where every task stands.

Getting started

New to gald3r? Start with the Quickstart - it walks through installing gald3r in a project and running your first commands.

Wiki sections

  • Quickstart - installation and first steps
  • Commands - every @g-* command, grouped by category
  • Skills - reusable capability modules
  • Agents - specialized agent personas
  • Rules - always-on conventions and enforcement gates
  • Hooks - lifecycle automation scripts

This wiki is the user-facing guide ("how to use gald3r"). For framework internals and contributor docs, see the source repository at https://github.com/Gald3r-Labs/gald3r.


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

This page is ingested one-way from the Gald3r-Labs/gald3r GitHub wiki. Edits happen on the wiki; this static copy regenerates on the next build — never the other way around.