mixedbit

5 ms·

1.
by mixedbit 15d ago
This assumes that a known watermarking algorithm is implemented. To me, unless output starts to include information whether watermark was inserted or not to a result, relaying on such an assumption is risky. If a human editor needed to hide
2.
by mixedbit 15d ago
I'm afraid watermarking could restrict applications where LLMs can be safely used to assist with writing. If I write something myself and use an LLM to proofread it, without watermarking I can confidently say that corrections done by L
3.
by mixedbit 20d ago
My experience with Boto: need some S3 manipulation logic, here is an official documentation that shows how to solve the exact problem with Boto, one caveat, this version of Boto is deprecated. Do the same with the newest version? Not possib
4.
by mixedbit 2mo ago
Apple stock is up 18% in 2026
5.
by mixedbit 2mo ago
It assembles and pivots to own root filesystem. The filesystem mounts dirs and files from your original filesystem, most of them in read-only mode, but does not mount files that expose secrets to the sandbox. The sandbox also mounts own, wr
6.
by mixedbit 2mo ago
I work on a sandbox which has similar isolation level to Podman (rootless Linux user namespaces), but with UX optimized for local development work. Take a look: https://github.com/wrr/drop Basically, you don't ent
7.
by mixedbit 3mo ago
Normal vending machine transactions are B2C transactions, so the buyer cannot be a company - cannot pay with company money and cannot deduce the payment as the company cost. I guess, the buyer can take a receipt from a vending machine and a
8.
by mixedbit 3mo ago
With payments the complexity is not only in accepting a payment, but largely in doing so legally. Someone makes a request to my company's paid service, I return 402 and get a stable coin back. Who do I invoice for this revenue? What va
9.
by mixedbit 3mo ago
You are right, but if stock repurchase does not increase market cap, dividend payment still decreases it (unless it would be fully reinvested in the same stock, which it is not, it is largely reinvested in largest caps).
10.
by mixedbit 3mo ago
Could outperformance of largest cap companies be partially explained by dividends paid by smaller caps? Reasoning behind this: * Index investing raises in popularity, with index funds that automatically reinvest dividends being often prefer
11.
by mixedbit 3mo ago
I work on a Linux sandbox that makes it easy to hide sensitive files from AI agents while keeping the files they need accessible. Check it out: https://github.com/wrr/drop
12.
/proc/self/exe overwrite from within a user namespace(mixedbit.org)
4 points by mixedbit 3mo ago|0 comments
13.
by mixedbit 3mo ago
Except for LLM services, which move away from subscriptions
14.
by mixedbit 3mo ago
I made a sandbox to productively work with agents while restricting files they can read and write: https://github.com/wrr/drop
15.
by mixedbit 4mo ago
one possibility is that some heavily indebted AI infrastructure company will be unable to meet its dept obligations, which will cause banks that become heavily exposed to AI-infrastructure related dept to tumble
16.
by mixedbit 4mo ago
sp500 has profitability requirement, I doubt LLM companies will show profits any time soon.
17.
by mixedbit 4mo ago
This is just an implementation detail, a feature flag can very well be implemented with a Boolean in a database. To me the main appeal of feature flags is that they allow to work on large features that often require months and many commits
18.
by mixedbit 5mo ago
When I was doing Fast.AI Deep Learning course, I was surprised by the number of Python dependencies machine learning projects bring. Web front-end projects were always considered very third-party dependencies heavy, but to me, the machine l
19.
by mixedbit 5mo ago
An interesting variant of a web phishing attack is to combine the back button hijacking with information that comes from the HTTP referer header. HTTP referer discloses from which website the user is coming from, when the user click the bac
20.
by mixedbit 5mo ago
I'm working on a sandbox for Linux terminal work with UX similar to Python virtualenv: https://github.com/wrr/drop
21.
by mixedbit 5mo ago
Recently I was wondering how viable it is to launch a niche, paid tool for Linux. I found that this is a very rare model, most tools are either just free, supported by sponsorship, supported by some paid cloud-based service that accompanies
22.
by mixedbit 5mo ago
OK, I see, so a limitation is also that I cannot block an individual script, I need to block a Python interpreter.
23.
by mixedbit 5mo ago
I'm not a Little Snitch or Open Snitch user, I wonder if these firewalls are able to block requests done with the use of some other, allow-listed program. Say I run a script `suspicious.py' and I deny this script from making any n
24.
by mixedbit 6mo ago
Check also https://github.com/wrr/drop which is a higher-level tool than bwrap. It allows you to make such isolated sandboxes with minimal configuration.
25.
Show HN: Sandbox agents without losing your dev environment(github.com)
4 points by mixedbit 6mo ago|0 comments
26.
by mixedbit 6mo ago
Is the OS implemented from scratch, or is it a stripped down version of some existing OS?
27.
by mixedbit 6mo ago
In Europe there have always been a good selection of affordable, entry-level vehicles, but recently these affordable choices have largely vanished. Toyota Corolla, the best-selling car in the world, is about 100% more expensive than pre-cov
28.
by mixedbit 6mo ago
I work on a sandboxing tool similarly based on an idea to point the user home dir to a separate location ( https://github.com/wrr/drop ). While I experimented with using overlayfs to isolate changes to the filesystem and
29.
Show HN: Drop – Linux sandboxing for LLM agents and untrusted code(github.com)
1 points by mixedbit 6mo ago|0 comments
30.
by mixedbit 6mo ago
If an author of a PR just generated code with an LLM, the GitHub PR becomes an incredibly inefficient interface between a repository owner and the LLM. A much better use of the owner time would be to interact with LLM directly instead of re

More ›