Hacker News Reader

Comment by mcoliver | Hacker News Reader

I have been happy with Google's Antigravity harness and Jules so looking forward to playing with this. Thanks for sharing. Simultaneously I am looking to also revisit local offline models.

While I feel like I have a decent understanding of the model landscape I'm feeling a bit lost at which agentic harness to leverage for local models. Hermes, Cline, Aider, Qwen Code, Goose, Pi, OpenCode, something else? I live in the terminal so Desktop UX is a bonus but not a must have.

Can I modify the antigravity settings/program to point to a local model? Where should I spend my energy?

View on HN

I'm stuck on Windows, so oh-my-pi has been really nice. The others I've tried such as kilo do alright but tool calling can mess up a bit.

Only complaint is that connecting the agent harness to my local model took more work getting configured right than I'd like, but that's been true of most harnesses I've tried as well. Most assume you're using a cloud model and local model configuration is a bit of an afterthought.

oh-my-pi has pretty poor permission system in my experience. Either yolo or deny/approve everything. No classifier, no sandbox.

Everything you want to add can be made as an extension, and pi has many ready extension to be added. For example, Sandbox: `https://github.com/earendil-works/gondolin` The sandbox extension is at https://github.com/earendil-works/pi/tree/main/packages/codi... And the extension at `https://github.com/earendil-works/pi/tree/main/packages/codi...` Which can also be found under `/@earendil-works/pi-coding-agent/examples/extensions/gondolin` at the npm modules

In case of omp, not sure if it's already at the node module package but you can just grab it from the links I shared and set it up.

I tried many of these pi extensions but they have tons of paper cuts. Don't remember which one had which issue but here are some I ran into:

* commands run by me (! prefix) are also sandbox blocked

* agent has no way to _request_ unsandboxed execution (e.g. if `kubectl whatever` is rejected by the sandbox, the model should have the chance to request permission)

* does not understand shell composition patterns (e.g. if `git status` is allowed and `git log` is allowed, then `git status && git log` should be allowed automatically)

* sandbox only supported on mac or linux. not both

All of that can be fixed by yourself. That's certainly the spirit of pi. But if you want strong defaults and batteries included (like omp promises) then that's just annoying.

My workflow is generally like so:

- instruct model to write a markdown file with a phased plan to implement whatever feature or change I want

- start a new context, instruct model to implement one phase of the file

- review changes manually, then start a new context and have it do the next phase

- repeat as needed

I've never seen omp touch a file outside of the directory I start it up in, and the few times where I've been unhappy with a change git has been there to revert.

This could easily be a case of survivor bias but I've not had an issue with letting it go yolo yet.

Sometimes agents create huge half-minified one-off python/bash scripts to do some data processing. I'd prefer to neither review nor yolo these. Sandbox restricts reads/writes to designated directories, so at least there's no `rm -rf /` in there.

What are you using Jules for? I want to like it but it fails too often. If I could use Gemini 3.8 I would be happy but 3.6 rarely succeeds.

Since local models are largest constrained by context window you want to have a tiny system prompt. I know Hax: https://github.com/OleksandrChekhovskyi/hax was designed with local models in mind, but I haven't used it.

i built something with a similar philosophy https://github.com/khimaros/hrns

Give https://juggler.studio a shot if you want a nice desktop UX with all the extensibility of things like Pi

> its core is open source

I'm not a fan of open-core apps.

Hmm, that phrasing is misleading - it's an open-source project: the core is AGPL, the extensions are permissively licensed

Deepseek Harness is great, at least with deepseek.

Does agy has auto mode?

Yes

This is news to me, do you have a link to docs for the configuration option?

goose has now native support for local models

Its docs have zero support

What's going on with Goose? Seems like Block donated it to some consortium; I can't tell if that's a good signal or a bad one. With so many "contenders", if Goose is going into maintenance mode it'd be helpful to know.

Goose is alive and well. It was indeed donated to Linux Foundation, and earlier this year the full-time team of core developers moved over to Spiral. The new focus is on GDK, goose development kit. Same great goose capabilities now being packaged for anyone to easily build their own customized agentic applications. See more here https://substack.com/home/post/p-205861523 and here https://goose-docs.ai/docs/gdk/sdk/

Did something change with it for depth of integration? Goose has been able to use local models via OpenAI compatible endpoints for at least a year

pi.

Always Pi.

Reply on news.ycombinator.com