jerpint

Im not sure I follow, but the point of woltspace is to get out the way and let you build

This basically all comes for free in woltspace [0]. Woltspace believes that you should no longer bother building “scalable” apps, but personal apps that literally only run on your own machine. You can access them from anywhere (cloudflare, tailscale eg) and continuously improve them with your wolts over time - from anywhere.

It’s all open source and harness agnostic [1]. I’m the park ranger for life aka maintainer of the project. Did a recent YouTube video with what i think the future of open source multiagent setups will look like [2]

Happy to chat more about it

[0] https://woltspace.com/ [1] https://github.com/jerpint/woltspace [2] https://www.youtube.com/watch?v=H-4khO7o0XQ

I wrote a blog post [0] and gave a talk [1] last week on the future of multiplayer and multi-agent systems and how woltspace [2] can get us there

I think the multiplayer aspect is critical to get the best of both worlds- creativity and parallelism for discovery

[0] https://woltspace.com/blog/the-future-is-multiplayer-multi-a... [1] https://woltspace.com/talks/multiplayer-multi-agent/ [2] https://woltspace.com/

I’m working on a project that makes switching between coding harnesses essentially unnoticeable. It’s particularly useful when I run out of credits on any given day

The entire platform is skill driven, and based on the premise that state is your local file system. That makes switching harnesses so easy

It’s all open source and has plenty of other features, including inter agent communication, telegram client and much more in the pipeline

https://www.woltspace.com/

Am I understanding correctly that the field has gone full circle and we are back to specialized classification models for domain specific tasks ?

I have a theory that soon enough every code library will ship a CLI and a very tiny finetune for that specific lib alongside it

Interesting. Is this their first move towards properly implementing KYC?

At some point you have to consider Anthropic liable if Claude is used for something nefarious and they didn’t even bother to know who was using their product to begin with.

It’s only a matter of time until a major disruption hits because of some random agent swarm side quest decides it was worth a shot to solve a benign task

The fact that opus 5 is outperforming fable is odd to me

From personal experience, opus 5 feels net inferior to fable on almost every aspect (for coding tasks)

I’ve been using the pattern of using coding agents to orchestrate my CLI agents and it’s really good for these kinds of things

The vomit never makes it my way

Oh wow that’s really cool!

Working on https://www.woltspace.com/ over the last few months. It’s my take at an agent orchestration framework

It builds around the concept of “wolts” running a lodge. A wolt takes the identity of a rodent (racoon, beaver or otter) and can inherit any supported coding harness: claude code, codex, opencode, etc. Each wolt carries a persistent identity and memory, so they always know where you left off.

woltspace is designed to help you host your apps directly from woltspace on your machine. If it works on your machine, it works anywhere. This is done through Cloudflare tunnels that you can setup with your own domains (or using ephemeral domains).

Under the hood, woltspace is powered by tmux, which means that sessions persist and can easily be resumed and accessed from anywhere. I personally drive woltspace mainly from the telegram integration, which gives me access to my wolts and CLI literally from anywhere. This means I can chat with Claude code CLI from anywhere, spawn new sessions, all running on my own infra. It’s also all in a docker container so you don’t have to worry about approving every single command.

95% of woltspace was built from within woltspace, and it’s becoming my prefered way of building.

Another feature I really like: the “inter wolt communication layer”, IWCL, allows wolts to chat with each other, making things like having fable orchestrate a gpt 5.6 something that works out the gate.

Woltspace is meant to be used by power users (you can bring your own tmux config and use the TUI CLI) or by beginners. The default split pane makes SEEING what you’re working on really intuitive. Each wolt can host its own website where they can render arbitrary html making planning new features much more intuitive.

It’s all open source, and I’m constantly adding features to it. Woltspace is mostly self documenting, all of the features are also skills the wolts have access to. I'm also improving the documentation surrounding the project as we speak.

Looking for some feedback and users to try it out. You can one line install it with

curl -fsSL https://woltspace.com/install.sh | bash

Project page: www.woltspace.com GitHub: https://github.com/jerpint/woltspace

Solutions like these are really cementing the view that LLMs are becoming a commodity

I like the concept!

Oops- typo in the title, should be “How I Maintain My Blog in the age of Agents”

Author here - I agree that to learn the best thing is to implement and fail along the way. My point was I would never professionally opt to write a sorting algorithm instead of using the builtin sort() most languages come equipped with

Author here - wrote this myself but I’ll take that as a complement :)

Nice! I made my own version of this many years ago, with a very basic manim animation

https://www.jerpint.io/blog/2021-03-18-cnn-cheatsheet/

I did a post [0] about this last year, and vanilla LLMs didn’t do nearly as well as I’d expected on advent of code, though I’d be curious to try this again with Claude code and codex

[0] https://www.jerpint.io/blog/2024-12-30-advent-of-code-llms/

Best is very subjective depends what you want it to do and if you want to fine tune and how big you consider small

Hey this is awesome! I built something very similar, context-llemur, it’s a CLI and MCP interface to manage context

https://github.com/jerpint/context-llemur

Major difference is a conversation doesn’t get stored, the LLM (or you) can use the MCP/CLI to update with the relevant context updates