Start Here

Getting Started

Prefer the old long-form? See V1 archive.

Install the devkit (once per machine)

git clone https://github.com/amit-t/ai-devkit ~/Projects/Tools-Utilities/ai-devkit
cd ~/Projects/Tools-Utilities/ai-devkit
./install.zsh
source ~/.zshrc

Adds three globals: init.wb, join.wb, wb.upgrade. .dev / .cly variants force Devin / Claude.

Requirements

Stamp (initiator, often QA)

mkdir ~/workbenches/wb-example && cd ~/workbenches/wb-example
init.wb

Devin prompts for label, epics, repos, roles, Figma refs, optional MCPs, target org (defaults to your login). Creates a private repo from the template, clones service + automation repos into repos/, seeds project.conf + EPIC-PIPELINE.md, writes .github/CODEOWNERS with you listed, commits, pushes. Share the resulting URL.

Join (collaborator, often dev)

cd ~/workbenches
join.wb https://github.com/<your-org>/wb-example

Devin asks what extra repos you want cloned, appends to project.conf, adds you to CODEOWNERS, commits, pushes.

Pull template updates

cd ~/workbenches/wb-example
wb.upgrade

One-way sync of template_owned paths only (skills, scripts, CLAUDE.md, AGENTS.md, aliases.sh). Your PRDs, specs, BDDs untouched.

First session

cd ~/workbenches/wb-example
source aliases.sh    # add to ~/.zshrc for next time
claude .             # or: devin .

Then /epic-intake EPIC-001 to pull your first epic and start a PRD.