ATechGuy
ATechGuy on Hacker News
This is very useful. Thanks for sharing!
Posted by
ATechGuy on Thursday March 19, 2026 @03:51PM
from the ask-hn dept.
In the last couple of months, several new solutions for sandboxing AI agents have launched (microVMs, WASM runtimes, browser isolation, hardened tool containers, etc.). Curious to hear from people using them in production. Are they working as advertised, or are there still major tradeoffs around security, cost, and performance?Here's my list of sandboxing solutions launched in the last year...
Posted by
ATechGuy on Wednesday March 04, 2026 @05:26PM
from the ask-hn dept.
In the last couple of months, several new solutions for sandboxing AI agents have launched (microVMs, WASM runtimes, browser isolation, hardened tool containers, etc.). Curious to hear from people using them in production. Are they working as advertised, or are there still major tradeoffs around security, cost, and performance?Here's my list of sandboxing solutions launched in the last year...
Posted by
ATechGuy on Wednesday February 25, 2026 @03:51AM
from the ask-hn dept.
Lately we've been noticing a non-trivial amount of traffic in our logs that doesn't look like typical bots.Not the usual noisy crawlers or obvious scrapers. The behavior is different with fewer hits, more selective page access.Some of the user agents suggest AI crawlers, but some do not. How can we track these visitors?
Posted by
ATechGuy on Tuesday February 17, 2026 @03:06PM
from the ask-hn dept.
[Posting again in hopes of gaining further perspectives and insights]Do you still browse the web the same way you did (say 5 years ago) or you use AI to browse, summarize, etc. on websites? Is it about speed/efficiency, instant answers (gratification), or something else?I've noticed I'm less willing to read long articles now that I can get instant summaries. I'm curious to understand if AI is...
TBH, this looks like an LLM-assisted response.
And how are you going to define what ocaps/flows are needed when agent behavior is not defined?
Posted by
ATechGuy on Monday February 16, 2026 @05:20PM
from the ask-hn dept.
Do you still browse the web the same way you did (say 5 years ago) or you use AI to browse, summarize, etc. on websites? Is it about speed/efficiency, instant answers (gratification), or something else?I've noticed I'm less willing to read long articles now that I can get instant summaries. I'm curious to understand if AI is saving us time or is it training us to avoid effort.
Interesting. I'd appreciate an example. Thanks!
Would love to see performance numbers with nested virtualization, particularly that of IO-bound workloads.
> a website can expose functions like searchProducts(query, filters) or orderPrints(copies, page_size) with full parameter schemas
How would this not create backend load and abuse?
That's a good start I'd say, but I agree with you that detection is not trivial. I wonder if there's enough value in distinguishing between AI agents (with full browser) and humans. What use cases would it enable?
Yes, but how can you tell if it was an agent with full browser?
Posted by
ATechGuy on Thursday February 12, 2026 @07:16PM
from the venturebeat.com dept.
Thanks for your insights!
QQ: does uxwizz also show AI agents visiting websites? have you seen traffic from AI agents?
Posted by
ATechGuy on Wednesday February 11, 2026 @01:45PM
from the ask-hn dept.
There's a lot of talk about agentic systems that AI can take actions, run workflows, and operate autonomously. However, when I look at AI startups, most of them seem to be building dashboards: analytics panels, control screens, monitoring tools.If the goal is autonomous agents, why are we still shipping dashboards? Is this a trust issue, a sales issue, or something deeper about how companies...
Posted by
ATechGuy on Thursday February 05, 2026 @01:29PM
from the euractiv.com dept.
I agree. However, how to define these permissions when agent behavior is undefined?
I understand the concern. However, you can customize the profile (e.g., allowlist) to only allow network access to required domains. Also, looks like your sandboxing solution is Docker based, which uses VMs on a Mac machine, but will not use VMs on a Linux machine (weak security).
That's right. But they (E2B) rely on the underneath Cloud infra to achieve high scalability. Personally, I'm still not sure about the value they add on top of Cloud hosted VMs. GCP/AWS already offer huge discounts to startups, which should be enough for VM-based sandboxing of agents in the MVP phase.
To the commenters here: thanks for correcting me! So AWS is losing AI sandboxing market to GCP due to high cold start times of EC2...very interesting!
You can customize curr_dir_access_profile.sb to block access to network/fs/etc. Why is this not enough?
sandbox-exec -f curr_dir_access_profile.sb markdownlint
Curious to know what made you DIY this?
These are all wrappers around VMs. You could DIY these easily by using EC2/serverless/GCP SDKs.
> we offer secure cloud VMs that scale up to 100k concurrent instances or more.
High scalability and VM isolation is what the Cloud (GCP/AWS, that E2B runs on) offers.
> I feel more in control of what's being exposed and how
Makes complete sense. Thanks for your insights!