frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Best Open Source Offline AI Agent

2•Ms-J•1h ago
With the rise of AI agents, I've been looking for an agent that can use a combination of local models or models that are hosted on servers I control.

For the local models, the agent must be able to run completely offline. When using remote models, it must only send network traffic to my servers. No checking for "updates" or pulling any type of dependencies from elsewhere.

When trying Opencode, it was a failure.

See: https://github.com/Chetic/opencode-offline

"This is the offline fork of OpenCode. Upstream OpenCode requires internet access for several core functions: the web UI proxies all requests to app.opencode.ai, ripgrep and LSP servers (TypeScript, Python, C++, Rust) are downloaded on first use, and model metadata is fetched from models.dev. In air-gapped or restricted network environments, this means no web interface (HTTP 500 on every request), no file search, no code intelligence, and no model definitions."

The problem is this project isn't maintained to standards and still uses Opencodes base.

Thanks for any suggestions.

Comments

josefcub•1h ago
Try charmbracelet's crush, found here:

https://github.com/charmbracelet/crush

Crush is pretty new, but getting better all the time. It's written in Go, so no node hijinks to get it working. It works fine with my ollama or llama-server localhost endpoints, and I've used it to make up a couple of internal projects without any issues.

It does have internal telemetry and such (including updating its list of external models it can use) that can be turned off in the crush.json configuration file.

If you're on a Mac, you can install via homebrew or use the more traditional route via Github.

Ms-J•22m ago
Thanks for the recommendation, I took a look and maybe you can answer a few questions that I couldn't find a clear answer when doing some quick searching.

Regarding local models, can it use them? I found this discussion:

https://github.com/charmbracelet/crush/discussions/775

I didn't appreciate the meow maintainer's attitude converting it into a discussion and ignoring the issue even to this day.

"It does have internal telemetry and such (including updating its list of external models it can use) that can be turned off in the crush.json configuration file."

Is there a page or guide which explains the telemetry and any internet connected settings?

Forgot to add, I use Linux.

Archerlm•1h ago
From what I know of, Ollama works offline, but if your ram is only 8 GB and without a GPU, latency would be severely limiting. The closest local model I can suggest is LM Studio (no cloud dependencies, own model registry, etc.). If that doesn't work out for you then GPT4All, but its models are quite small.
Ms-J•20m ago
I'm looking for an agent, but thanks.

I forget the issue with Gpt4all as some have blended together when they weren't suitable for me.

Show HN: Proposal for a real long-term AI memory benchmark

https://penfieldlabs.substack.com/p/proposal-a-new-benchmark-for-long
1•dial481•1m ago•0 comments

Little Snitch comes to Linux, but the core logic is closed source

https://the.unknown-universe.co.uk/privacy-security/little-snitch-linux/
1•TheIPW•2m ago•0 comments

Referi is quietly becoming the system behind referrals

https://www.referi.net
1•onlinemelvin•4m ago•1 comments

Show HN: Mdpdf a 2k line C CLI to convert Markdown to tiny PDFs

https://github.com/schicho/mdpdf
1•bastscho•5m ago•0 comments

Show HN: BrokenClaw Part 5: GPT-5.4 Edition (Prompt Injection)

https://veganmosfet.codeberg.page/posts/2026-04-08-openclaw_gpt5_4/
2•veganmosfet•6m ago•0 comments

Agents Need a Database to Break

https://xata.io/blog/every-ai-agent-needs-a-database-it-can-break
1•gk1•7m ago•0 comments

Nice Email Service

https://beeble.com
2•janwebdev•8m ago•0 comments

Show HN: LLM-Wiki but for Early Founders

https://github.com/himanshusaleria/gtm-engine
4•ashed96•11m ago•2 comments

Linux CFS Quotas and Throttling, explained as a children's story

https://www.mrpengi.com
2•perarneng•13m ago•0 comments

Show HN: Browse museum collections tik tok style

https://artsnap.vaguespac.es/
2•thebigship•14m ago•0 comments

Ask HN: How to get people to use your product?

2•pbs29•15m ago•0 comments

What the Verdict Against Meta and Google Says About the Way We Live Now

https://www.newyorker.com/news/the-lede/what-the-verdict-against-meta-and-google-says-about-the-w...
1•Brajeshwar•15m ago•0 comments

"There is something wrong with me" negative beliefs associated with dissociation

https://www.sciencedirect.com/science/article/abs/pii/S0022395626001500
1•rendx•16m ago•0 comments

Great resources for self-hosting AI hardware

https://github.com/voipmonitor/rtx6kpro
2•chisleu•17m ago•1 comments

Open-source AI assistant that watches your screen and points at things

https://github.com/farzaa/clicky
1•steveharing1•18m ago•0 comments

Show HN: We made 25 years of a NYC gallery's exhibitions conversational with AI

https://contxt.art/bitforms
1•scriptedfantasy•19m ago•0 comments

Show HN: Job-focused lists of product companies using Go, Rust, Scala, Elixir

https://readytotouch.com/
1•harakternik•19m ago•0 comments

Designerd

https://chromewebstore.google.com/detail/designerd/lhlfjbdaffoimeokmfifciickpmckmjf
1•rgb_dsgn•21m ago•0 comments

Dead man's switch for your crypto keys and passwords

https://github.com/bonkai/afterkey
3•bonkaiii•21m ago•0 comments

Coding Agents Are Reading Your .env

https://infisical.com/blog/your-ai-coding-agent-is-reading-your-env-file
6•jakehulberg•22m ago•0 comments

Pwning NetBSD-Aarch64

https://www.feyrer.de/NetBSD/bx/blosxom.cgi/nb_20260308_1932.html
1•t-3•22m ago•0 comments

Claude Cowork generally available on all paid plans

https://claude.com/blog/cowork-for-enterprise
2•aghuang•24m ago•0 comments

Nyth AI – Local LLM Inference on iOS Using MLC-LLM and TVM

https://apps.apple.com/us/app/nyth-ai/id6757325119
1•rida123•24m ago•0 comments

Petri, a multi-agent orchestration framework for building AI context

https://github.com/onthemarkdata/petri
1•onthemarkdata•24m ago•1 comments

Rechunk: Adaptive, Embedding-Only Retrieval

https://github.com/chessstetson/rechunk
4•chessstetson•25m ago•1 comments

I open sourced KubeezCut a browser-based video editor (editor.kubeez.com)

https://editor.kubeez.com/
2•Meeeeeep•25m ago•1 comments

Show HN: Render Workflows – Durable task orchestration without queues or workers

https://render.com/workflows
4•anurag•27m ago•0 comments

Building Custom ASICs – Podcast

https://theamphour.com/721-chip-design-for-fun-and-waffles-with-julia-desmazes/
1•random__duck•28m ago•0 comments

Foxhound – helping LLMs better navigate codebases

https://github.com/karimfan/foxhound
1•karimfan•29m ago•1 comments

Show HN: Shellbeats just a pure C music streaming from terminal

https://shellbeats.com/
1•lalospace•29m ago•0 comments