corv

6 ms·

31.
by corv 8mo ago
When their questionnaire asked me for feedback I specifically mentioned that I hoped they would not reduce visibility to the point of Github Actions. I guess that fell on deaf ears.
32.
by corv 8mo ago
I’ve heard this before that common doses are unnecessarily high but why is that? Patents?
33.
by corv 9mo ago
Really goes to say something about starving artists
34.
by corv 9mo ago
I happen to use a Mac, even when targeting Linux so I'd have to use a container or VM anyways. It's nice how lightweight bubblewrap would be however. Consider one wanted to replicate the human-approval workflow that most agent har
35.
by corv 9mo ago
I like the bubblewrap approach, it just happens to be Linux-only unfortunately. And once privileges are dropped for a process it doesn't appear to be possible to reinstate them.
36.
by corv 9mo ago
You're right that blocking on every operation would defeat the purpose! Shannot is able to auto-approve safe operations for this reason (e.g. read-only, immutable) So the agent can freely explore, check logs, list files, inspect servic
37.
by corv 9mo ago
Thank you, good to know it landed :)
38.
by corv 9mo ago
I'm pursuing a different approach: instead of isolating where Claude runs, intercept what it wants to do. Shannot[0] captures intent before execution. Scripts run in a PyPy sandbox that intercepts all system calls - commands and file w
39.
by corv 9mo ago
Looks like the underdog beats it handily and easier deployment to boot. What's the catch?
40.
by corv 9mo ago
I’m building Shannot, a human-in-the-loop sandbox for AI agents on production systems. Instead of filtering commands with heuristics (which agents work around), it dry-runs entire scripts in a PyPy sandbox, captures every command and file o
41.
A field guide to sandboxes for AI(luiscardoso.dev)
3 points by corv 9mo ago|0 comments
42.
by corv 9mo ago
The gist dismisses sandbox-2 as “might as well use Docker or VMs” but IMO that misses what makes it interesting. The PyPy sandbox isn’t just isolation, it’s syscall interception with a controller in the loop. I’ve been building on that foun
43.
by corv 9mo ago
Great documentation of the problem! The bypasses logged all stem from the same root problem: policy sandboxes give agents constraints to optimize against. I’ve been exploring a different model: capture intent instead of blocking actions. Sc
44.
by corv 9mo ago
I’ve been working on a different approach to this problem: syscall-level interception via PyPy sandbox rather than command filtering. This captures all operations at the OS level, so tmp.sh scripts and Makefile edits get queued for human re
45.
by corv 11mo ago
When AI and robots take care of everything there is more time to make babbies
46.
by corv 11mo ago
Absolutely. I’ll send you an email
47.
by corv 11mo ago
Interesting! The sandboxing space definitely deserves more attention. On the other side of the spectrum, we're working on a lightweight approach that augments user namespaces with libseccomp to filter syscalls via BPF. https:/&#x
48.
by corv 1y ago
We've been exploring libseccomp/BPF filters on top of bubblewrap namespaces for LLM sandboxing - the BPF layer lets you go beyond read-only mounts to syscall restrictions. Open to collaboration on pushing this further: https:&#x2
49.
by corv 1y ago
I've replaced my OrbStack usage entirely with Podman Desktop and have zero issues with it, unlike with OrbStack. In particular the 1TB VM disk image OrbStack uses wreaks havok with deduplicating backups. Their disk cache also caused me
50.
by corv 1y ago
Seeing Treeform here immediately made me think of Nim, and lo-and behold that's what the Javascript is generated from, cool!
51.
by corv 1y ago
Happy to see it get more attention here - it really is versatile
52.
by corv 1y ago
Awesome and in Nim!
53.
by corv 1y ago
> There’s also a Russian torpedo that uses a similar concept to “fly” at supersonic speeds underwater. Supercavitation
54.
by corv 1y ago
The missing magic incantation to create your Brewfile: brew bundle dump
55.
by corv 1y ago
It's really good but the fs caching has bit me a couple times before I realized what was going on.
56.
by corv 2y ago
Adieu Mozilla!
57.
by corv 2y ago
I’m glad to see M5Stack is supported!
58.
by corv 2y ago
This is great to hear but I’ve found Kagi’s Orion to be so good that I’m unlikely to switch
59.
by corv 2y ago
You can map ChatGPT to the action button and talk to it from the lock screen.
60.
by corv 3y ago
Another way of going about it would be to require transparent algorithms. Any mass media platform could be problematic regardless of who owns it.

More ›