Just found a nice patch of sun in a comfy spot on the couch. Maybe cats do have some things worked out.
Any conversation
@stagasprofilefollow
creative technologist
creator of textlog, editing on mobile breaks the “280 chars / 10 lines max use and @mentionsand more”. Writing a new note doesn’t.
creative technologist
creator of textlog, editing on mobile breaks the “280 chars / 10 lines max use and @mentionsand more”. Writing a new note doesn’t.Really glad to be using Haskell for work - especially in today's climate with agentic coding.
Our team uses containers to keep our development environment consistent across team members, though, and most agent sandbox approaches want to either offload your work fully to the cloud or they want to sandbox the agent harness process itself.
We're not ready for full cloud based development (yet?) and sandboxing just the agent doesn't work for us when the agent needs to run docker.
... because letting the agent run docker gives them a too-easy escape hatch from their sandbox.
So we're currently stuck with using VMs. This turns out to be tricky but seems do-able.
My current approach is to provision a VM image and use Incus for the execution. My main goal is to keep my current workflow of local/non-agent dev working seamlessly and share the source with the VM/agent.
This means sharing my working dirs with the agent VM but overlaying artifact dirs from VM-owned block devices so we don't take too big of an IO perf hit.
I also don't want the agent to be able to
git push so that means we need some quirky git config to rewrite origin paths and thread through some agent specific PATs ("thanks" GitHub) for read-only repo access. These ALSO need to be threaded through to the development containers - they need to clone private repos.... in the end, this seems to work but I haven't lived with it for long.
There are still some quality of life issues to work out, but I think it'll work out.
At this point, I've removed all coding agents from my host - and that's nice.
Curious, why does using Haskell make a difference here?
It's tangential to the sandboxing topic, really. I started the first post with a broader intent in mind and then elaborated on the sandboxing setup I'm working on.
For AI-assisted coding, though, it's nice because Haskell provides a lot of feedback earlier in the development process compared to lots of other languages. It tends to work once it compiles (obviously not all the time).
I find myself building bespoke tooling for my own and workflows these days, not only because of the obvious reason of software being so much cheaper to build, but also because there are SO MANY options for solving any given problem that it feels impossible to evaluate all of them.
If it's not a crucial piece of tech for a work problem then it often feels like paving my own way is both cheaper mentally (kind of amazing that's true these days) and probably a better fit.
It will become increasingly harder to find niche markets for software when those people can just build it themselves exactly how they want it. In the end, people want to accomplish tasks, and the AI is going to be building the software on-the-fly exactly for that task, why do we need generic software that may or may not be a fit?
Yeah. It makes me think that protocols for exchanging data between apps will become even more important in that landscape. But maybe, not - maybe each exchange is another bespoke AI-driven implementation?
Protocols, infrastructure, containers, those kinds of things. I was actually playing around with an idea for a minimal markup language + runtime that can be utilized by AI to spin up apps more easily, and brainstorming it with ChatGPT, I have a feeling something like that will appear soon to complement MCP.
Routines in Claude Desktop create Claude Code sessions, so hourly routine litters in iPhone' Claude.app Code tab. Screw it, built my local version of AWS StepFunctions - Stepper - and run skills as state machines there.