DevForge

Context

DevForge started from a recurring frustration: every time I set up a new machine — whether a fresh install, a new laptop, or a VM — I would spend 3 to 6 hours reinstalling tools, configuring shells, downloading language runtimes, and remembering which aliases I depended on. The process was not just slow; it was inconsistent. Each setup would drift slightly from the last, creating subtle bugs that wasted even more time.

Why I Did Not Use Ansible or Chef

The obvious choice for machine automation is a configuration management tool like Ansible, Chef, or Puppet. I deliberately chose not to use them. The reason: they solve the problem of managing fleets of servers, not the problem of setting up a single developer machine. Ansible requires Python on the target machine, a playbook to maintain, and a SSH setup that feels heavy for a personal laptop. Instead, I wrote pure Shell scripts. They have zero dependencies, run on any Unix system, and are trivial to inspect and modify. The trade-off is that Shell scripts are less declarative than Ansible playbooks, but for a single-machine setup, the simplicity gain is worth the verbosity loss.

The 200+ Tools Problem

The second design challenge was scale. A modern developer environment is not just a language runtime — it is homebrew, git, docker, node, python, go, rust, neovim, tmux, fzf, ripgrep, lazygit, and 190+ other tools. Installing them all sequentially takes hours. The solution was parallel installation groups: tools are categorized into groups (languages, CLIs, editors, AI tools, docker), and each group installs in its own background process. A status bar shows progress for each group. Total installation time dropped from 3 hours to ~45 minutes on a fast connection.

AI-First Design

What makes DevForge different from other dotfile repos is the AI-first philosophy. It installs and configures: LLM CLI tools (llm, aichat), local model runners (ollama with pre-configured models), AI-enhanced shell completions, and IDE extensions for GitHub Copilot, Continue.dev, and Supermaven. The idea is that a machine in 2025 should ship with AI capabilities ready to use, not as an optional afterthought.

Template Architecture

DevForge is designed as a template repository, not a framework. You fork it, edit the `profile.sh` file with your preferences, and run `./forge.sh`. This means: no package manager dependency, no npm install, no database. The entire project is less than 500 lines of Shell. Every function is documented with a header comment explaining what it installs, why it is useful, and what dependencies it expects.

Gallery