20+ years of coding experience across video games, logistics, finance, real estate, and legal/law industries. I'm a versatile, abstract-thinking solver of problems across many domains and disciplines
I've never taken the time to "pay the tax" of environment setup. For better or worse, I just pip install globally and go. 😅
I come from the C++ and C# world, where you download, install, and just get coding. Python, for as much as I love the language, always seems like it has this "tax" as you put it. The pageantry. That said, it's awesome to see this sort of feature. I'm probably the kind of person that would find it super useful. 😅
Something that popped into my head while reading this one - my Chaos Monkey cannot be contained apparently - how much time or thought have you put into a bad actor using your VMs for malicious activities?
I imagine there's not much that could be done while maintaining the utility of your project, but it's worth considering. At the very least you'll be aware of some attack vectors so you're not blindsided later.
David, this is exactly the kind of "Chaos Monkey" thinking that keeps a platform honest. Thank you for bringing it up!
(And you are spot on about Python's "pageantry"—coming from the C++/C# world, it definitely feels like extra steps!)
To answer your question: preventing malicious activity is a constant balancing act, but we have several hard architectural guardrails in place to ensure the blast radius of any bad actor is effectively zero:
Hardware-Level Isolation: We don't use shared Docker containers. Every sandbox is an isolated E2B Firecracker MicroVM. They are stripped down, ephemeral, and have no persistent storage between sessions.
Strict Resource Quotas: Every MicroVM has hard cgroup limits on CPU and RAM. If a bad actor tries to spin up a crypto miner or a resource-exhaustion script, they hit the ceiling instantly and the process is throttled or killed by the kernel.
Aggressive TTL (Time-To-Live): Sandboxes are ephemeral by design. They auto-destroy after a short, predefined period. There is no time to establish a persistent foothold.
Auth & Rate Limiting: Every deployment is tied to a verified Supabase Auth account. We enforce strict concurrent sandbox limits per user. If an account exhibits abusive behavior, it is instantly banned, and all its active sandboxes are forcefully terminated.
It’s definitely a cat-and-mouse game, but these layers ensure that the utility for legitimate developers remains high, while the platform is inherently hostile to sustained abuse.
Thanks again for keeping me on my toes and helping me think through these edge cases! 🐯🛡️
AI testing storyteller. Writing the 36 Stratagems series — ancient Chinese military tactics meet modern AI system meltdowns. 16yr QA → building AI test framework.
AI testing storyteller. Writing the 36 Stratagems series — ancient Chinese military tactics meet modern AI system meltdowns. 16yr QA → building AI test framework.
AI testing storyteller. Writing the 36 Stratagems series — ancient Chinese military tactics meet modern AI system meltdowns. 16yr QA → building AI test framework.
@effessdev , @akhourianmolkumar , @hubedav , @nyaomaru , @xulingfeng GUYS?
Keep building! 😸
Thanks bro
Ohh 😮 you forget me!
Sorry oops 🤦♂😅
I don't know what to say bro 🤔
I've never taken the time to "pay the tax" of environment setup. For better or worse, I just
pip installglobally and go. 😅I come from the C++ and C# world, where you download, install, and just get coding. Python, for as much as I love the language, always seems like it has this "tax" as you put it. The pageantry. That said, it's awesome to see this sort of feature. I'm probably the kind of person that would find it super useful. 😅
Something that popped into my head while reading this one - my Chaos Monkey cannot be contained apparently - how much time or thought have you put into a bad actor using your VMs for malicious activities?
I imagine there's not much that could be done while maintaining the utility of your project, but it's worth considering. At the very least you'll be aware of some attack vectors so you're not blindsided later.
David, this is exactly the kind of "Chaos Monkey" thinking that keeps a platform honest. Thank you for bringing it up!
(And you are spot on about Python's "pageantry"—coming from the C++/C# world, it definitely feels like extra steps!)
To answer your question: preventing malicious activity is a constant balancing act, but we have several hard architectural guardrails in place to ensure the blast radius of any bad actor is effectively zero:
It’s definitely a cat-and-mouse game, but these layers ensure that the utility for legitimate developers remains high, while the platform is inherently hostile to sustained abuse.
Thanks again for keeping me on my toes and helping me think through these edge cases! 🐯🛡️
@effessdev , @akhourianmolkumar , @hubedav , @nyaomaru , @xulingfeng GUYS WHAT YOU THINK?
Love the idea of cutting that annoying setup tax! Excited to see where this project goes 😄
Lol test it and poke holes lol i love to see fellow devs holes and feedbacks 🤣
Hahaha I’m pretty lazy myself! I’ll give it a shot sometime when I feel like it. I just love lying around doing nothing 😂
Lol
🤣hahaha
Hahahahaha 🤣