Why I switched from iTerm to MOLTamp

I used iTerm2 for about six years. I had it customized within an inch of its life — keybindings, profiles for different projects, status bar with git branch and battery and CPU, custom prompt, the works. It was the most personalized piece of software I owned. And I switched to MOLTamp anyway. Here is the actual reason, which is more interesting than the obvious "MOLTamp looks cooler" version.

iTerm is a great terminal emulator

I want to be clear about what I am NOT saying. iTerm2 is excellent. The keybinding system is mature. Split panes work beautifully. The profile management is industrial-grade. The shell integration features (badges, marks, semantic history) are deeper than anything else on macOS. If your goal is "the best possible TERMINAL EMULATOR for macOS," iTerm2 is still the right answer and I would not argue otherwise.

The problem is that I no longer want a terminal emulator. I want something different.

What changed

What changed is that I started spending most of my terminal time inside Claude Code. And Claude Code, as a category of tool, is not "command-line program." It is "agent runtime." The terminal stops being a place where I type commands and gets output, and starts being a place where I have a conversation with an AI that is editing files, running commands, making decisions, and asking my permission.

The ergonomics for that workflow are different. The visual layer matters more. The status bar matters less (because the agent is doing things in the background and the relevant signals are about agent state, not shell state). The profile system matters less (because I am not jumping between dozens of project shells anymore — I am parked in one project, talking to one agent). The keyboard customization matters less (because I am typing prompts, not commands).

iTerm2 was optimized for the wrong thing. Not the wrong thing in 2018. The wrong thing for what my actual work has become.

What MOLTamp is optimized for

MOLTamp is optimized for "I am about to spend several hours in conversation with an AI agent and I want the room to feel right." Not in an aspirational way. In a literal way. The features that get the most attention are:

Skins with real personality (not just color schemes — full visual identities)
Widgets that surface non-shell information (time, weather, music, calendar, system stats, ASCII pets)
A vibes deck for art / GIFs / mood
Audio visualizers
Layout management for "focused" vs "browse" vs "presenting" modes
Per-skin customization that does not require editing dotfiles

None of those features are about typing commands faster. All of them are about being in the room longer.

What I gave up

Honestly, less than I expected. The iTerm features I was sure I would miss:

Profiles per project — turns out I never actually used these. I had 15 of them set up, and I used the same 2-3 by reflex. Replaced with "open MOLTamp in this directory, adjust nothing."
Split panes — still useful but I use them less now, because the right sidebar widgets cover most of "the second thing I want to look at." When I do need a real second shell I open a tab.
Custom prompt with git status etc — replaced by the git widget in the sidebar, which actually shows MORE information than my prompt did, without taking up any horizontal space in the terminal area.
Keybindings I had memorized — I had to relearn a few of these, but the keybinding count dropped because I am typing prompts now, not chaining commands.

The thing I miss most is iTerm's shell integration "marks" feature, which let you jump back to previous prompts. Not implemented in MOLTamp yet. I have learned to live without it.

What I gained

This is the more honest part of the post. The things that genuinely changed for the better:

I want to be at my desk more. This is the big one. My iTerm setup was optimized but not delightful. MOLTamp is delightful and the difference is measurable in how often I sit down to start work.
The visual feedback for agent state is real. Skins react to what the agent is doing — color shifts when an agent is thinking, tool-call indicators, visible permission requests. iTerm has none of this because iTerm does not know what an agent is.
Music and visualizers are integrated, not adjacent. The act of having to maintain a separate music app and a separate terminal had a real friction cost I did not realize until it was gone.
Skinning is actually easy. I built three custom skins in my first week. I had used iTerm for six years and never built one custom theme — the friction was just high enough that I never started.
The community gallery makes other people's work accessible. I download skins and widgets from people I have never met and they just work. iTerm has community profiles but they live in random GitHub repos and require manual installation.

Should you switch

Honest answer: only if your workflow has shifted toward AI-agent-first the way mine has.

If you spend most of your terminal time in raw shells running commands, building, debugging traditional code without an AI, iTerm2 is still better for you. It will always be better at being a terminal emulator than MOLTamp will be, because iTerm has an eight-year head start on the shell integration features.

If you spend most of your terminal time in Claude Code, Codex CLI, Gemini CLI, or Aider — having conversations with an agent that runs your tools for you — then your needs have changed even if you did not notice. Try MOLTamp for a week. The first two days will feel weird because you keep reaching for iTerm features that do not exist. By day five you will have found your new defaults. By day ten you will not want to go back.

The fact that your work has changed does not mean iTerm2 is bad. It means iTerm2 is for an earlier phase of computing. MOLTamp is for the phase you are probably in now.