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.
Any conversation
... 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).
(Opinion) People donโt like generative AI because it made the {insert creativity} too easy. Saw the HN post on โAI music banned from top 40 chart in Australiaโ and people had very strong opinions in both directions.
The hot take: If youโre ok with someoneโs creativity being outsourced to others, then you should be ok with AI doing the same.
If โAI made this and I call it mineโ is bad but โsomeone overseas on fivver made this and I call it mineโ and itโs suddenly ok, time to self reflect.
โฆ
My hot take: from the moment you use technology, be it an instrument, a synthesizer, a DAW, and now AI, it is still your idea, your effort, your time spent, and that makes it yours. Others can do the same if they choose. The same way someone chooses a classic guitar over a synthesizer, one can choose an AI over (insert ancient way of doing things).
โฆ
Youโre rolling the dice whenever you are searching for a sample or a preset in a synthesizer. Eventually itโs what โsounds goodโ, so still the effort is your judgement and the time spent finding that โdice rollโ that sounded well.
I am not a musician, so I lack the jargon to properly explain what I mean.
I used AI for creative writing (nothing too fancy, TTRPG). There, the broader the dice roll, the more generic and uninteresting the results. Slop, if you will. It was surprisingly good to build up on ideas I had however.
I imagine muisic also has building blocks. Melodies, instruments, mixing, etc. AI can perhaps be used in a similar fashion there.
I heard a perspective I sort of like: if AI is a tool then acceptable AI use is, more or less, defined as "when my audience no longer notices or cares". It's dynamic like that. Bad photoshop is detectable and annoying, so are overusing presets. Your artistry needs to expand to meet the tool.
I don't know if I 100% like that or agree. At the same time, it's a pretty straightforward line, a practical take.
we also launched Norma! :)
It checks AI-generated code for security, quality and maintainability while the agent is still working, rather than after a pull request opens. The output is deterministic. Here the demo link: portal.qualityclouds.ai/demo
@fastidiousprofilefollow
"Beware those that would deny you access to information, for in their heart they dream themselves your master." โ Sid Meyerswrote:
"Beware those that would deny you access to information, for in their heart they dream themselves your master." โ Sid Meyerswrote:I had hit a throttle now. Thatโs a โsubtle" way to tell me to slow down.
๐๐ฎ๐ฐ ๐ ๐๐ฎ๐ฐ ๐ช๐ฏ ยท๐๐ฎ๐ฆ๐๐ฆ๐ ๐๐ฆ๐๐ฉ๐ ๐๐ฆ๐ ๐ข๐ฐ๐๐ง๐ฏ๐: ๐ฃ๐จ๐ ๐ฉ ๐๐ซ๐ค ยท๐ฆ๐๐๐ค๐ฆ๐ ๐ฏ ๐ฉ ๐๐ฒ (๐ฎ๐จ๐๐ฆ๐) ๐๐จ๐๐๐ผ๐๐ฑ, ๐ฏ๐ฌ ๐ ๐๐ต ๐จ ๐๐ฐ๐ ๐ฎ๐ด๐๐. ๐ค๐ณ๐๐ค๐ฐ ๐ฏ ๐๐ท๐๐ฐ ๐ฆ๐ฏ ๐๐ฆ๐ ๐ท๐๐ณ๐ฅ๐ฏ๐ฉ๐ค ๐ข๐จ๐๐ผ.
Three for three on British dishes this weekend: had a full English and a pie (rabbit) yesterday, now to do a beef roast. Lovely and cozy in this autumnal weather.