That is the value. You can run them remotely from your phone and set them up on a headless machine too. You’re not bound to your terminal.

My usual workflow is to still use the terminal for heavy work. But sometimes I just want to prompt from my phone and run a few tasks or do some planning. Then I can simply resume the session in the terminal later. It's the flexibility.

If you’re happy with OpenCode as the harness, OpenChamber is great. But if you prefer using different harnesses under the hood, Paseo is a better fit.

I installed it on my homelab and can access the same sessions from my MacBook or iPhone at any time. Been really nice to use so far.

I originally built this to scratch my own itch while job hunting.

The main problem for me with the monthly Who Is Hiring thread was finding relevant jobs by location. I was mainly looking around Europe, and most posts are understandably US focused.

I’ve since found a new job, so I don’t really need it myself anymore, but I thought it might still be useful to others.

The crawler is written in Go, but the final site is just a single HTML file plus JSON files, served through Cloudflare Pages. The crawler runs every 10 minutes on an Ubuntu VM on my existing Synology homelab.

The extraction/classification part was also a bit of an experiment for me.

I primarily use Sol at work, but I also had a GLM subscription that I picked up during their Christmas deal last year, so I wanted to see how well it handled both coding and this kind of structured classification task.

Right now the extraction is done using sequential GLM 5.2 calls with reasoning disabled. I tried batching multiple job posts together as well, but found I needed reasoning enabled to reliably avoid cross-contamination between posts. Unfortunately it is noticeably slower.

I also tested 5.6 Luna medium through pi and got broadly similar results. GLM has been slightly faster for this particular workload so I’ve kept it for now.

I have a small golden dataset to test the parser/extraction against, although there will definitely still be some false positives since I don’t manually verify every job.

Most of the project itself was vibe coded using GLM 5.2 through Claude Code using Paseo, which also meant I could work on it from my iPhone.

It’s free, no account required.

Hopefully it’s useful to someone. Happy to consider feature requests, and examples where the parser gets something wrong are especially useful.

Comments URL: https://news.ycombinator.com/item?id=49230662

Points: 3

# Comments: 1

These things often look obvious and simple after the fact. Finding the weakness in the first place is the hard part, and that’s what makes the agent’s capabilities interesting here, especially at scale.

I’m asking because I’ve been looking for agent harness comparison tools too. I’m interested in more than just the inputs and outputs—I also want the system prompts, traces, and tool calls. It’s useful to understand why Codex, for example, uses more tokens while Pi doesn’t.

Fewer tokens aren’t necessarily better if the agent skipped important checks. On the other hand, using more tokens could just mean it’s overthinking the process. Either way, seeing the full execution trace for the same task is really valuable.

I spend most of my time in the terminal and have been using tmux to manage my sessions and workspaces, mainly via sesh.

But recently, I decided to move to Kitty and use its built-in sessions instead. I built something around it as well. It’s still early, but I’m really liking this workflow so far.

https://github.com/alienxp03/kesh

Comments URL: https://news.ycombinator.com/item?id=48829851

Points: 2

# Comments: 1

I don’t think this should be framed as iOS vs Android. From a technical perspective, I understand why the behavior differs. But from a user’s perspective, all they see is that one app works much better than the other.

My family doesn’t care about the technical reasons. They just saw one app finish much sooner than the other, and they preferred the one that worked better. Based on that experience, I can’t recommend Immich to them.

After looking through it, the database design was a mess. Some features worked, some didn’t. I explained the missing pieces and why things were breaking. Like OP said, he’s the domain expert.

I used billions of tokens last month alone. The tools are getting better fast. But giving AI to a domain expert doesn’t mean you no longer need software engineers.

A domain expert can use AI to build software. And a software engineer can use AI to learn about the domain. Both bring different expertise to the table.