HackerTrans
TopNewTrendsCommentsPastAskShowJobs

Bullhorn9268

no profile record

Submissions

[untitled]

1 points·by Bullhorn9268·2 miesiące temu·0 comments

Estimating the market value of top AI researchers

futuresearch.ai
11 points·by Bullhorn9268·3 miesiące temu·0 comments

LiteLLM PyPI has been compromised an hour ago, do not update

futuresearch.ai
27 points·by Bullhorn9268·4 miesiące temu·5 comments

Caution: Read the Docs for Claude 4.6's Effort Parameter

everyrow.io
8 points·by Bullhorn9268·4 miesiące temu·2 comments

comments

Bullhorn9268
·4 miesiące temu·discuss
The fact pypi reacted so quickly and quarantined the package in like 30 minutes after the report is pretty great!
Bullhorn9268
·4 miesiące temu·discuss
Not a security researcher, but this is IMHO obviously positive that the other side of the arms race is also getting stronger, and I would argue it's stronger than on the bad guys' side, due to the best being somewhat responsible and adding guardrails.

I like the presentation <3.
Bullhorn9268
·4 miesiące temu·discuss
I am from futuresearch and went through this with Callum (the OG). We did a small analysis here: https://futuresearch.ai/blog/litellm-hack-were-you-one-of-th... of the packages and also build this mini tool to analyze the likelihood of you getting pwned through this: https://futuresearch.ai/tools/litellm-checker/
Bullhorn9268
·4 miesiące temu·discuss
yeah, updated in the post
Bullhorn9268
·4 miesiące temu·discuss
So, is there a way of somehow cleaning up those docker containers, e.g. when uv version is not available?
Bullhorn9268
·4 miesiące temu·discuss
I believe cloudflare tries to work around that with Code Mode, right? https://blog.cloudflare.com/code-mode/ (but the OP is still useful for when classic MCP is needed)
Bullhorn9268
·4 miesiące temu·discuss
I see the value in making it simple for the user, but here I feel it's a bit too much. Would probably prefer two.
Bullhorn9268
·4 miesiące temu·discuss
I too learned this the hard way. I still haven't figured out why this is the case - like, is the inherent incidental irreducible complexity of human life too high, so it dominates majority of our actions?
Bullhorn9268
·4 miesiące temu·discuss
I dislike spotify but weirdly enough I can never point out exactly what's so frustrating about it...
Bullhorn9268
·4 miesiące temu·discuss
Weirdly, CC still adds stupid comments to guide its next steps despite having that instruction in CLAUDE.md AFAIK.
Bullhorn9268
·4 miesiące temu·discuss
Maybe like that the humans would not properly review the more complicated ones too? I don't think it's a big danger to be clear.
Bullhorn9268
·4 miesiące temu·discuss
Some of the rules are impossible to code generally enough, right? Like, "no defensive coding" - I can't imagine what kind of AST shenaningas you would have to do without basically banning `if`.
Bullhorn9268
·4 miesiące temu·discuss
I mean, have you tried to ask it not to do those checks? It's similar to the OP about no defensive coding. In general, CC works better when you can tell it WHAT TO DO as opposed what NOT to do, but sometimes there is no way around it. What kind of checks it's adding? Like asserting that columns exist and stuff?