I've been biasing towards VMs myself just out of caution, but maybe that's just extreme paranoia.
The way I look at it is similar to how I'd look at any hypothetical employee. How do I ensure the agent only has access to the minimum possible they need to get their job done?
That means no access to git repositories (no pushes on my behalf means it can't accidentally nuke git history, something there is anecdotal evidence of agents doing). It can make local changes in git only and I will take responsibility for pushing them. No access to the wider internet beyond what I deem acceptable. No permissions to access any internal APIs except what I provide (and not using my credentials).
In one case, I have a tool that has a set of dangerous commands alongside a large number of safe ones. I don't even have it installed in the agent's VM. I run an MCP that is a simple python wrapper around the tool on the host side, and expose it to the agent in the VM, so that it can only possibly run a strict safe subset that I can trust it with access.
The easiest, most guaranteed way to isolate it is to run it in a VM or container where it literally can't do the wrong thing without some kind of full container or VM exit exploit.
It's not hard, it's trivial. Most folks here are constantly working with containers. You know how to run a container with a local directory mounted in it.
For myself, I've been using Lima (https://lima-vm.io/) to reduce even that little bit of extra work. Lima works cross-platform leveraging native virtualisation or containerisation, and has some useful capabilities for using agents.
Agreed. In my current job, across several large java code bases that have been developed over the last 12 years, the only place I've seen xml come into play is for maven. We've hundreds or more similar scale code bases and I don't think I've ever seen it in use beyond that, though I only occasionally look at those code bases.
No one is ever reaching for xml or even thinking about it.
> I still consider it some of the best multiplayer experiences I've ever had though.
I played Eve for a few years as part of a corporation in Xetic and then Ascendant Frontier.
So many painful large battles (time dilation got added after I stopped playing), and some wild solo fights. My favourite was the time I got caught solo in a T2 Interceptor, when out scouting. We knew an attack was coming but didn't know where.
I screwed up, and found myself surrounded by 5 enemy player ships, with no possibility of escaping. The only thing going for me was that I was in an inty, and they were in larger ships, so I could outmanouver them. I knew I was done for. If I flew away they'd be able to hit me as the only thing keeping me safe was my radial velocity (I was orbiting the ship faster than their weapons can rotate, but that only really works 1 on 1, to the other ships you're not moving quite as fast)
It was really just about how long I could hold out and making sure I was ready to warp the moment I got podded. I constantly switched orbit between ships, trying to keep them close together so I could maintain high radial velocity, while taking pot shots at them and starting to chip away at armour, and taking glancing shots from them myself. It felt like that fight went on for hours, but it was probably only 5 or so minutes before they finally managed to pod me, and I managed to warp away to freedom. That was probably nearly 20 years ago (I stopped playing maybe late 2007 / early 2008?) and I still remember it vividly. Once I'd got myself to safety I remember just sitting in my seat staring at the screen, as the adrenaline faded.
Back around 10 years ago, I gained a new manager who had previously managed mechanical turk. It was already recognised as a dead end back then.
I remember him talking about getting a mandate from Amazon Security to upgrade from the long EOL MySQL 4.0 to MySQL 5.something, and that it was almost impossible to get any resources committed from leadership to even do it despite the fact it was security requiring it (which usually resulted in everyone jumping before stopping to ask how high to jump). I want to say he ended up doing it himself? Something like that..
All existing extremely minimal headcount was tied up in a massive technical debt of KTLO work, and proposals to resolve those issues similarly met resourcing road-blocks.
> I have never seen a company use AMD outside of wafer and a couple others mostly in US.
There's a few using them, and even more starting to experiment with them. AMD has long been a source of disappointment around this side of things, so I'm hesitant to feel optimistic we'll finally get some competition. The market really needs viable competition to Nvidia, especially performance/watt.
Not sure why you're getting downvoted, this is the entire point of open source.
Does such a bug exist in Windows? OSX? Who checks? If someone finds the key in memory, can they tell what conditions might be causing it and where?
Their only recourse under those situations is to hand it off to the OS Vendor and trust that what they implement does solve the problem, and trust that it wasn't a deliberate back-door that is now being replaced by another back-door.
When I moved to the states nearly 20 years ago, my pronunciation sharpened up (stopped dropping consonants) very quickly. Over time I stopped using idioms, and the few bits of Cockney Rhyming slang I used (Butchers, Scooby; which are look, and clue, respectively).
I think it was less as a conscious act and more as a result of just not being around people that use them. There's a sizeable element of cultural reinforcement involved.
That said, they'll pry my British spelling out of my cold, dead, hands.
I don't think this is where you were going with your comment, but I'll mention this just because you're somewhat adjacent to a routine mistake in business:
Uber is a people delivery company, but they've had a lot of bright engineers working for them on their infrastructure and software over the years, and that work has rippled out through the industry.
Amazon (in VMWare's words) is "a company that sells books", and their leadership couldn't accept they were losing to them ("I look at this audience, and I look at VMware and the brand reputation we have in the enterprise, and I find it really hard to believe that we cannot collectively beat a company that sells books.").
NUMA is one of those amazing things that trip you up in all sorts of ways at unexpected times. The amazing "invisible" performance killler (invisible because unless you're already aware of NUMA, or remember to check, you won't know it's there potentially crippling you.)
It has been a source of routine conversations with customers and engineers of all kinds, and often one of those things you don't know about until too late.
I don't know if the kernel has improved this behaviour in the several years since last tested, but a coworker realised that the linux page-cache wasn't fully split by NUMA node. They were benchmarking mysql running it in each NUMA node, and noticed the second NUMA node was noticeably slower. Then discover after a reboot the second node was fast, and the first was slower. After a bit of thinking and tinkering they discovered that libmysql was ending up in the page cache in the same NUMA node as the benchmark client was run in first, so even though they were pinning the benchmark tool and mysql process to the NUMA node, the benchmark client was causing the OS to reach across the NUMA node to get at the page cached library.
Intel suffers just as much when NUMA enters the picture, even prior to CCD style architecture. That extra latency hop across to the other core to get at memory is absolutely crippling, especially in a hot loop. It requires very careful handling, while being this kind of invisible element (unless you know to look for it, nothing will draw your attention to it)
In theory having it in the same language could provide performance benefits, vs reaching out to a C library and dealing with C interop fun. Any language that is JIT'd would have native bytecode to optimise instead of a black box.
There are software engineers that specialise in source code analysis that lawyers will often use in these cases. The engineers will be given access to source code in secure environments where they're not allowed to bring any device in or out. They review, analyse, and write up a report using pen and paper, that can then be reviewed by the lawyers.
Working for an ISP in the mid 00s, the lack of security of those scripts was an absolute nightmare. It was a routine task to have to go clean up the mess they made, everything from simple "Being used to relay spam" on up.
That takes me back a bit. It's a perl-ism. I used to think it was a great design feature but I've come to strongly prefer "There should be one way to do it, and it should be obvious"
A few years back I heard (via a friend who knows a CEO, so take it with appropriate pinch of salt given not a direct experience), that Gartner was asking the CEOs in its group meetings how many of them were laying off people due to AI, with the question presented in a very leading fashion that implied anyone not was behind the curve and failing their business or whatever.
I've been a sysadmin for decades, dealt with *nix based servers since the late 90s, yet for the most part I've used devices like Synology servers, simply because I don't want to have to manage technology to that degree at home.
I've built my own NAS when my last synology died, and I'm not sure I'll build one again. I've dealt with all sorts of issues that I just haven't had to deal with with a packaged solution, and I really just want to not think about that stuff when I'm not working.
Yes, I can absolutely do it for cheaper, better, and with more flexibility myself. Doesn't mean I actually want to.
Gmail used to offer a low bandwidth / performance webmail interface, that was essentially their original UI. Ran like greased lightning, used barely any memory. Emails loaded almost instantly.
Kids of varying ages that I've spoken to often talk about the environmental impact (mind you, I live in a fairly liberal/left leaning part of the country), among other things.
At the risk of over generalising, I mostly hear a lot of shit talk from younger generations, distrust from millennials, and more excitement and interest from Gen-x-ish and older.
As with many things, there's a certain level of hypocrisy to the shit talking, because teachers are at the schools are complaining to parents about the kid's use of AI, and pointing out that they will automatically fail any writing that seems to be using AI.
The way I look at it is similar to how I'd look at any hypothetical employee. How do I ensure the agent only has access to the minimum possible they need to get their job done?
That means no access to git repositories (no pushes on my behalf means it can't accidentally nuke git history, something there is anecdotal evidence of agents doing). It can make local changes in git only and I will take responsibility for pushing them. No access to the wider internet beyond what I deem acceptable. No permissions to access any internal APIs except what I provide (and not using my credentials).
In one case, I have a tool that has a set of dangerous commands alongside a large number of safe ones. I don't even have it installed in the agent's VM. I run an MCP that is a simple python wrapper around the tool on the host side, and expose it to the agent in the VM, so that it can only possibly run a strict safe subset that I can trust it with access.