hobofan

18 ms·

331.
by hobofan 10mo ago
I'm not really into malware, so I was just wondering: - Isn't this really non-viable in practice? The "few headers" that were shown include an Authorization header, that would presumable rotate every ~24 hours and would
332.
by hobofan 10mo ago
I think for the next 20-30 years it still wouldn't be feasible to make it a red flag due to a non-negligable part of the older population not being on social media at all. I assume it is right now still be possible to enter countries w
333.
by hobofan 10mo ago
> Several times I've had to find someone young (and who speaks English) and say, "I'm just going to follow you to the next train." I'm on the other side of this. I end up chaperoning lost tourists on my DB disast
334.
by hobofan 10mo ago
Depending on what regions/lines you are exposed to one can definitely land on different sides of "it works well in general" vs. "it doesn't work well in general". Of course if you cite a Sprinter line like Nure
335.
by hobofan 10mo ago
> persistent and public I don't think that it's actually public? From one of their explainers, no public IP is assigned, so you'll need to ar least have to use an additional service like Cloudflare Tunnel to use it for hos
336.
by hobofan 10mo ago
Clay as in clay.com. But yeah, it shouldn't be assumed that people know that.
337.
by hobofan 10mo ago
> Aside: Why are LLMs good at greenfield? I have the complete opposite experience, where once some patterns already exist 2-3 times in the codebase, the LLMs start to accurately replicating them instead of trying to solve everything as o
338.
by hobofan 10mo ago
So does Cursor in the Plan mode.
339.
by hobofan 10mo ago
If you totally alter the character model to fit the envelope of a Boston Dynamics Spot/similar "dog robot', sure.
340.
by hobofan 10mo ago
Yeah, I foresee a bite of '27.
341.
by hobofan 10mo ago
Yeah, same here. For the comments it took into account it made pretty great roasts, but would have been better if it was actually comprehensive over the course of the year.
342.
by hobofan 10mo ago
I also think its fine and fair to charge for the general GHA infrastructure that one would also be using with self-hosted runners. I suspect that they weren't looking to make money off of those charges, but rather use that as a forcing
343.
by hobofan 10mo ago
Anecdotally I've seen it get a lot more common to use third-party managed runners (e.g. Blacksmith) for anyone that needs slightly beefier machines and/or a caching system that actually works.
344.
by hobofan 10mo ago
Why do you think a software engineer or computer scientist would be more qualified?
345.
by hobofan 10mo ago
MCP-UI and OpenAI Apps are converging into the MCP Apps extension specification: https://blog.modelcontextprotocol.io/posts/2025-11-21-mcp-ap...
346.
by hobofan 10mo ago
Whether you need a secondary set of release notes is almost purely a "marketing" decision, and should as much as possible be treated as such. Doing that is also usually way more effort (e.g. producing screenshots, etc.) than just
347.
by hobofan 11mo ago
> The Gemini API will randomly fail about 1% of the time, retry logic is basically mandatory. That is sadly true across the board for AI inference API providers. OpenAI and Anthropic API stability usually suffers around launch events. Az
348.
by hobofan 11mo ago
Contrary to what a lot of the other comments here are claiming, I don't think that's the mark of death for MCP and Anthropic trying to get rid of it. From the announcement and keeping up with the RFCs for MCP, it's pretty obv
349.
by hobofan 11mo ago
They haven't really. One of their latest blog posts is about how to retrofit the "skills" approach to MCP[0], which makes sense, as the "skills" approach doesn't itself come with solutions for dynamic tool disc
350.
by hobofan 11mo ago
Most of the value lies in its differentiation to OpenAPI and the conventions it brings. By providing an MCP endpoint you signify "we made the API self-describing enough to be usable by AI agents". Most existing OpenAPI specs out t
351.
by hobofan 11mo ago
It's not like there aren't already military AI startups in the EU. e.g. Helsing.
352.
by hobofan 11mo ago
At least as of now Germany has a robust enough social safety net and decent path for non-university careers that make a "poverty draft" system as it exists in the US not viable. On top of that there is a large dislike in the socie
353.
by hobofan 11mo ago
SeaweedFS has been our choice as a replacement for both local development and usage in our semi-ephemeral testing k8s cluster (both for its S3 interface). The switch went very smooth. I can't really say anything about advanced features
354.
by hobofan 11mo ago
Even outside of code development, Anthropic seems to be very strongly leaning into code interpreter over native tool calling for advancing agentic LLM abilities (e.g. their "skills" approach). Given that those necessitate a runtim
355.
by hobofan 11mo ago
> with a high degree of certainty No it didn't. As the top comment in that thread points out, there were a large number of false positives.
356.
by hobofan 11mo ago
So, why exactly is this one month old ad on the frontpage?
357.
by hobofan 11mo ago
Software is eating the world. AI is eating the VCs.
358.
by hobofan 11mo ago
One doesn't exclude the other. You can use machine translation and clearly marking it as such, and then let people override them if there is the necessary contribution activity.
359.
by hobofan 11mo ago
> This really depends on company size It mainly depends on whether the company is trying to sell to governments that have accessibility requirements anchored into their procurement process requirements. Of course there is some causality
360.
by hobofan 11mo ago
> So you are ok with 2FA, right? Yes. Are you not? It's one of the most effective measures to prevent a whole class of supply chain attacks. On Github the 2FA is also flexible enough to allow non-hardware passkeys, so you can choose

More ›