Quickstart

Install the signed binary

One signed Go binary per machine — no Python, no clone. Windows and macOS installers are signed as Gald3r Labs LLC; Linux ships a tarball.

# Windows — download gald3r-windows-x86_64.msi from
# github.com/Gald3r-Labs/gald3r_core/releases, run it, open a new terminal, then:
gald3r --version

# macOS — download gald3r-macos-arm64.pkg from the same releases page, double-click, then:
gald3r --version

# Linux
curl -LO https://github.com/Gald3r-Labs/gald3r_core/releases/latest/download/gald3r-linux-x86_64.tar.gz
tar xzf gald3r-linux-x86_64.tar.gz && cd gald3r-*-linux-x86_64 && ./install.sh

Check the install, then scaffold your project

Run these from your project root. The doctor never fabricates a green result; the dry run shows what setup would write before it writes anything.

gald3r doctor
gald3r setup --dry-run
gald3r setup

Project it into your IDE (optional)

The same commands become native slash commands in your AI coding tool — /g-task-add in Claude Code, @g-task-add in Cursor.

gald3r platform install claude-code   # or cursor, codex, copilot, ...

Confirm everything loaded with gald3r status.

gald3r status

Create your first task

Pick something small and concrete — a bug fix, a new endpoint, a UI component. Pass it as an argument.

@g-task-new Fix the broken 404 page styling

Gald3r creates .gald3r/tasks/task001_fix_404_page_styling.md with acceptance criteria. Review it — edit the criteria if needed.

Implement the task

Run the implementation command. The agent reads the task, implements it, checks each acceptance criterion, and marks the task [🔍].

@g-go-code

Verify in a new session

Open a new chat window (new agent session). Run the review command. This is the adversarial gate — a structurally separate agent checks the implementation.

@g-go-review

If all criteria pass: the task is marked [✅] complete. If something fails: it goes back to pending with a FAIL note, and you repeat from step 5.