we've been building a new @trylle experience bringing your code more close to your agents, all in one place. sharing more soon!

when you run agents for long inside sandboxes, it sucks if it stops or crashes and you lose your work! so we fixed that I believe boat.dev now has the most robust filesystem snapshot system out there: - snapshots include all files, installs, rights, users, browser sessions & in-VM docker state - they happen constantly, no need to request, literally can't forget to do them - you can fork a VM while it's still running - you can bake a VM image this way that you can reuse later for all your tasks or sub-users - free latest snapshots per VM: the snapshot sizes are super small, therefore cost is low. + you don't have to pay to keep them up - 2x backup in 2 EU countries for extra safety - diff-based, incremental, never saturates you or us, always space & resources efficient, minimum failure risk - even works seamlessly with DBs (even if in a Docker container) that you're running inside the VMs - lazily restored behind FUSE for fast resumes/fork, files are instantly usable but might initially load slowly, then disk becomes local nvme speed again - still not the fastest thing ever possible but we're improving every week opus and I made an explainer video, using blender + ffmpeg + headless chrome, all rendering scenes in parallel across a fleet of boat VMs:

my GitHub Actions spend is going out of control

I just posted this on YC's internal forum (bookface) but I think it's important that any AI founder reads this otherwise you're about to take a very bad vendor decision

we just shipped Magic Sessions in Trylle a new way to run longer coding tasks directly from your repos try it on trylle.com - start from an issue - kick off a task - test the change - open the PR here’s me taking a real issue all the way to a pull request