frenchie4111

First, this my favorite book of all time so I love seeing it shouted out

Second, I agree with exactly this. Something shaped like an AirPod

idk man fingerprints don't taste that good

Yes I want push to talk in ness

But trying to implement it directly into ness hasn't worked that well. It seems like betterwispr/wisprflow/etc all seem to belong better as their own app, and we just need good integration points between the push-to-talk app and the text consuming apps (like the IDE)

This is awesome. I really want to get something like this integrated directly into my agent IDE (https://getness.dev).

The main thing I need is some way of auto-sending a "end of message" button (like a newline would do) at the end of the message so that it auto-sends the chat.

Not sure the exact right way to implement that, but it seems like something that could be general purpose. Maybe a setting per-active app?

I noticed earlier this week that my URL bar now pre-fills githubstatus.com instead of github.com when I type "gith"

It's a step in the wrong direction but also token efficiency has become a focus relatively recently (just the past few weeks it seems like the zeitgeist has turned it's attention to efficiency) while work on this model probably started many many months ago. I would expect to see models released that focus on token efficiency in 6-12 months

Agreed - my intention is definitely not "harness on every server" and more "make it easy for the backend half of harness to run on your personal dev box". This just comes from my experience at BigTechCorp where all of the engineers (myself included) used beefy EC2 instances for development (so tests ran fast, lol) and VSCode remote to actually edit files

While I agree it is predominantly text getting sent over the wire, it would be pretty complex to implement harness on just a text stream without the websocket abstraction (I thought about it for a while). If you want to just SSH and run an agent you can, but the point of harness is that it has better ergonomics than a bare terminal can provide

When I built harness IDE one of the core goals was to stop having to leave my laptop open all the time while my agents finished the refactor.

Launching today is harness' SSH bootstrap mode. It works very similarly to VSCode's "connect to remote" feature. Select SSH connection you already have configured, it will connect, install the harness server in ~/.harness, and then the UI will connect over websocket.

Would love feedback if anyone gets a chance to try it out

This is awesome! Some feedback - I can't tell anywhere from the website if there is mobile support (which is a must-have if I want to consider moving my company or friends over to this)

Totally in agreement with you and the previous commenter. It's not a full re-write. I took an open source project (DocuSeal) and customized it for our needs

That said - Docusign has a moat because people are afraid (and CYA) by using them instead of anything else. To make an eSignature legal isn't really that complex, you just need hash based attribution, some timestamps and some ToS.

Replacing Docusign with something else is no where close to having an AI represent you in court

Agreed - They are definitely doing this price change ahead of the Steam Machine on purpose, to anchor that price better.

That said there is hope for the Steam Machine to at least be similarly priced. I'd expect the COGS to come in cheaper than the deck because it doesn't have a screen, battery, joysticks etc and it's likely easier to assemble in general

If I didn't love my Steam Deck I would consider re-selling it for profit. This feels like the same thing that happened to used car resale prices during covid. For a year or two you could sell the used car you just bought for a profit

This happened to me recently with a few providers. Their billing teams got aggressive and I said fuck it and AI coded a replacement in a few nights/weekends. For me it was: Temporal, Zapier and DocuSign

I strongly believe the build/buy equation is much different in 2026 than it was in 2024

Hi - I built Harness [1] as a reaction to the closed source IDEs that keep popping up (Antigravity/Conductor/Cursor). I wanted something better than those that is also community supported.

Harness is an agent first IDE built around git worktrees. It's designed to make it easy to organize a bunch of Claude Code sessions for all the different projects you are tackling right now

It supports macOS and Linux. And experimentally it allows separating your frontend & backend so that the agent runtimes can live in a dev box while the UI can run on your laptop.

Let me know what you think

[1] https://harness.mikelyons.org/

Agreed - On Harness I am already working on triggering/review. Those are the two next projects on the list

We will definitely continue to iterate / try to keep up with whatever everyone thinks is the best UX

Remote dev box is brand new in the last release so let me know if it works... happy to help you set it up (I personally run w/remote dev box) shoot me an email

And headless server support! I have gotten a lot of request to let people run the agent on a separate dev box (in the cloud or some pc gathering dust in the corner). The frontend/backend are decoupled so you can run the agents on a totally separate machine than the GUI is running.

If you have any other problems/concerns/issues/suggestions/etc, reach out!

Wow - this looks like a fork of my project https://github.com/frenchie4111/harness

It's crazy to see how we have independently landed in the same place

Good luck to your project! Excited to see how our projects differ in the future

Looks like Pi does do it (I was not an aware of Pi before now). It's obvious the industry is standardizing around git worktrees + agents, it's just about which tool has your favorite ergonomics at this point.

For me I liked the ergonomics of a few other tools, but none of them were exactly what I wanted so I made my own. And, I kept it open source so anyone can tweak the ergonomics to be what they like

I regret naming it harness for this exact reason but I am too deep at this point

I am building an Agent IDE called Harness. It is somewhat inspired by the previous version of antigravity (and Conductor, and a few others). But with a core goal being open source & hackability.

It's centered around git worktrees. The goal is to organize all your AI sessions into somewhat logical places and make it easy to context switch. The secondary goal is to remove the need to open a separate code editor anytime you want to look at a file (We have a built in file editor powered by Monaco [vscodes editor])

Check it out https://harness.mikelyons.org

I get that they have to make changes to the google search box because so many people are just using ChatGPT/Cluade to answer questions instead of google.

However, I specifically use Google (or DDG) when the LLMs are failing me. When I want "research something on my own" because the LLM is giving me garbage, or untrustworthy information. If Google completely replaces their search box my Google usage will go down even further.

I don't plan to use Google's LLM when Cluade is just better. Now that Google's search features are gone (or going away) I no longer have any reason to turn to them at all

Totally agreed, and that's exactly why I built harness! If you use just raw worktrees you can get lost super quickly because there's just things happening all over the place

Harness solves this by putting the status of the agent front and center (or rather on the left vertical tab bar). Each worktree is a tab, the icon shows you if the agent is working (green means working, yellow means waiting on you, red means waiting on a permission prompt). The goal is that all work for a worktree is in one mental space, so you can easily review the agent's work (reviews built in, integrated browsers sorted by worktree, etc)

And of course harness [1] (I kid, this is my tiny personal project in the same space)

[1] https://harness.mikelyons.org

Does anyone use Zed with a monorepo?

I am in a situation where every sub-folder has its own language server settings, lint settings, etc. VSCode (and forks) can handle this by creating a workspace, adding each folder to the workspace, and having a separate .vscode per-folder. I haven't figured out how to do the same with Zed.

I would love to stop using VSCode forks

Github's migration to Azure has so far been a hilariously bad advertisement for Azure

wow you guys really delved into this

I found this part interesting: "Inference requests from the agent never leave the sandbox directly. OpenShell intercepts every call and routes it to the NVIDIA cloud provider."

Seems like they are doing this to become the default compute provider for the easiest way to set up OpenClaw. If it works out, it could drive a decent amount of consumer inference revenue their way

lol - it's literally called malus but I guess that's only an obvious giveaway in retrospect