Moving Away from OpenForge, for Now

I am pausing OpenForge development for now. Not because I stopped wanting the workflow. I want to use it without having to keep maintaining the entire app around it.

The part I still value is simple: fast, keyboard-first ticket triage and a clear view of which tickets need my attention. OpenForge does that well. Moving away means giving up some UX I genuinely prefer.

Adding features was the easy part

AI made it easy to add features to OpenForge. Polishing them was harder.

Getting the abstractions right and fixing the accumulating papercuts became difficult to keep up with. Being able to add another feature did not make the existing ones easier to maintain.

That is the distinction I keep coming back to. Building the workflow I wanted and maintaining a whole application turned out to be different commitments. I still care about the first. I do not want to keep taking on the second right now.

My usage has also changed. I am doing more from my phone, and a terminal experience on mobile is tricky. The workflow needs to fit how I actually use it, not just how I use it at a keyboard.

A better base, with tradeoffs

I tried Herdr too, but ran into limitations. In bb, I have built everything I wanted and more. The integration has been smooth and saved me a lot of headaches.

That does not mean I prefer every part of it. Ticket navigation in bb is worse than OpenForge’s today. Its current interface is mouse-first, and I still miss having a terminal. The fast, keyboard-first triage I like in OpenForge is still what I want.

What appeals to me about bb is its plugin architecture, its ability to work across multiple machines, and its mobile support. I see those as a stable foundation. I can improve my workflow without owning the entire app underneath it.

An interface I can steer from

I also think a richer editor or interface can help me see context, steer agents, and get better results. That is a belief about how I want to work, not proof that more UI makes an agent better. I want to see where it is going while I can still change its direction.

Peter Steinberger, the creator of OpenClaw, describes a preference for seeing where agents are going and steering early and often.

Loading X post. View post on X

His criticism of an agentic UI without steering or message queuing gets at the same concern. That resonates with me: I want an interface that helps me participate, not just wait for an answer.

Loading X post. View post on X

I take launches like Meta Muse, Grok Bot, and OpenAI’s dots as signs that personal agents and agents we work alongside will become commonplace. I expect other companies to follow. That is my prediction, not a settled outcome. It makes me more interested in building my workflow on a foundation I can extend than maintaining another whole app.

Still experimenting

bb’s community is not the biggest yet. I am also trying OpenClaw because of its growing capabilities and larger community.

That is an experiment, not a conclusion. I do not yet have a concrete outcome to point to, and I am not treating it as a proven replacement. I am still figuring out what fits.

The decision I have made is narrower: I am pausing OpenForge development. I am not declaring it dead or ruling out coming back to it.

The workflow is worth keeping. Maintaining the whole app is not what I want to keep doing right now.