Last time I tried Windows though, coding agents were pretty finicky on native Windows for some reason. Maybe this gap is closed now though!
If you have the tokens to spend, please give it a try. I am curious to see how well it will work. The toolchains it uses, Bun & Zig, do work there, and I think most of the system tricks have Windows equivalents
I haven't used OMP before. Let me take a look :)
I put a lot of small touches like that throughout the app. It just makes me feel cozy inside haha.
I can see that it closes one vector of attack (attacking the install script serving), but beyond that the user still runs the binary outside a sandbox.
I feel that the binaries themselves, both directly and indirectly (supply chain attack) are the elephant in the room.
Any thoughts on that? I always wished the cli stuff I install from Homebrew was locked down by default, but it's not practical with today's computer systems since it kills usability.
Anyway, big fan of Homebrew. Looks like I can't submit a formula to core until I get more traction on GitHub (stars/forks/watchers). Otherwise I have to go to taps which takes away a lot of the benefits of being on homebrew-core iiuc.
I work off of several machines, and the lack of native remoting has been frustrating for me so it’s the next big thing to add :)
Can you share the theme clunkiness you’re seeing? I tried to avoid having another theme system for users to config, but maybe these adaptive tricks won’t quite cut it.
My email is [email protected]
The huge ts files are indeed concerning, but in practice are more resilient than I had expected. They do eventually break down though, and I have been refactoring the systems as they reach their limits. (the last one was a breaking up agent providers into modules, motivated by wanting to support Pi and OpenCode)
I do have some very basic git tooling in there but would be awesome to beef it up :)
I’ll take a look then, as I am unsure how sandboxing will interact with this system since it does a few unusual things
It works ~alright in light themes, but not as well considered as the default one yet which is why it's not the default.
Can you give it a try?
Maybe I am missing something, but the second you run the code, it has the same powers and risk profile as a bash curled from the internet. Maybe something about having explicit versioning?
(as an aside, I think this will change very soon…)
Hope you like it!
For the website, it would be helpful to have either a screenshot or a "sandbox" thing so I can play with it a bit
Hey everyone,Inspired by Herdr, I've built an agent multiplexer that I've been using as my daily driver for a few weeks now.I think I improved on the concept in a few ways, focusing on the ergonomics I always wished I had:I show you what the agents are up to, along with a running feed of their activity. There's also a simple shortcut (F1) to go to the next unread agent.I use multiple accounts,...
If the goal is keeping people informed, one can show various breakdowns, piecharts or any other creative infographic at the canvas of the bill.
A common one sees the restaurant charge you separately for the employee health benefits, cynically marketing it to consumers as “healthy SF” and “SF health mandate”.
What’s sort of fun is that like cryptocurrencies, you can always materialize new ones out of thin air. One popular establishment demands that you pay a separate “carbon offset fee”.
Something I am nervous about is using a @hey.com email. Similarly to domain, it's a world-visible and sticky, but unlike domains, it isn't portable, the only options are: pay Basecamp forever, or lose access. gulp
I often wish the language had adopted it instead of the C#-like await async, since it’s just more straightforward.
Anecdotally, I couldn't get an interview, even though I had an offer from them in the past.
Anecdotally, I once checked into an Airbnb with a leaky roof. After spending a good 20 minutes trying to get to a human (a small eternity when you’re in an unfortunate situation), Airbnb asked that I spend hours negotiating with my non-responsive host on a solution, which is obviously nonsensical given the situation.
Being able to instantly talk to someone and feel like I am being taken care of would have made all the difference in the world. Instead, I churned from their platform.
Did you end up using this as a way to estimate how "healthy" the agents are, or was this explored after the system was already working well?
HotelTonight has been awesome there. You can get a representative on live chat and resolve most issues while you're still en-route from the airport.
On one hand, I am hopeful Airbnb will be able to learn from HotelTonight, but in reality, it seems most acquisitions end up with the acquired company diffusing into nothingness :(
It is awful.
The bikes: big, heavy and ugly. Due to their sheer weight, peddling up the gentlest of hills will make you question your life choices, and consider other, much faster options, like walking, or just gently slithering around the pavement.
Additionally, Motivate's chosen method of bike cleaning - a dude hosing them down - tends to leave the bikes drenched with ice cold water. I suppose the feeling of freeze gnawing at your body does act as a replacement for caffeine for some, but I find it very unwelcoming.
The stations: Sparse, not well maintained, and honestly, an eyesore. They're this giant, lumbering object that takes away precious public resources, and is only available for Motivate's usage. You also constantly have to plan around them -- is there one next to where I want to go? Is it almost full?
E-Bike deployment: what a joke. They are rarely available, and even you've spotted one of these precious unicorns, you'll often by greeted by a beaming red dot, indicating that the bike is unwell, likely due to the battery giving up in the middle of day.
Other than that, the app spams you with ads, many of which are strangely enough for buying an e-bike. I thought this company is about bike sharing??
Looking forward to playing the updated version on the Switch!
It's like how in Uber/Lyft, most people just default to 1/5 star ratings. In that case, the average rating a driver has to maintain gets skewed to a a pretty high number, like 4.5, and people who think "oh, this was a pretty good ride! 4 stars!" end up unintentionally boning the driver.
Is there a way to only view commits that belong in the current branch (like git log)? The commits from the other branches are distracting when you're trying to read the history of a particular branch.
In general, the whole thing feels rushed and duct taped together. Networking model (inherited from Docker) doesn't play nice with ELB.
- The built-in AWS tools for monitoring are not container aware.
- We've had multiple occurrences of the ECS daemon dying.
- Very little visibility into the progress of deploys. The API/console will report something as "running", when it in fact was still loading up.
If you watch their videos, they promise integration with Marathon, but if you look at the code, it's in "proof of concept" stage.
At this stage, GCE is significantly ahead of AWS. Out of the box, you get a top notch container story, logging and monitoring.
I do wish it had a global symbol search, similar to Sublime's (cmd+shift+r). It's indispensable for code navigation.
It's literally a tutorial on configuring Alexa Voice Services + their sample code on Debian.
The way you interact with it is by clicking on a button in a Java app. No trigger phrase like Echo.
I want to use this, but I don't want to give you full access to my bank account.