devmor

15 ms·

271.
by devmor 6mo ago
That too! Every time the agent does something I didn't intend, I end up making a tool or process guidance to prevent it from happening again. Not just add "don't do that" to the context.
272.
by devmor 6mo ago
I have to work with agents as a part of my job and the very first thing I did when writing MCP tools for my workflow was to ensure they were read only or had a deterministic, hardcoded stopgap that evaluates the output. I do not understand
273.
by devmor 6mo ago
Throughout my life, what colleagues/friends are unwilling to remark plainly on has been the most telling factor of someone’s character to me.
274.
by devmor 6mo ago
No, most systems in daily life can be understood if you are willing to take the time. That doesn’t mean you personally are required to, but some people do and your interaction with the system of social trust determines how much of that rema
275.
by devmor 6mo ago
It is very silly to treat zero grounding the same as accepting core, proven concepts. Your PoV here is no different than saying "It rains because god is sad and crying" is an appropriate thing to believe. If you want to say "
276.
by devmor 6mo ago
I would say I'm using it for about half of the "10%" and and a quarter of the "90%". > I can't imagine that slopping it up would be a great decision. Having alien code that no one ever understood between a b
277.
by devmor 6mo ago
I was attempting to help someone who runs a small shop selling restored clothing set up a gemini pipeline that would restage images she took of clothing items with bad lighting, backgrounds, etc. Basically anything that showed any “skin” on
278.
by devmor 6mo ago
Copilot isn’t really a competing product to Claude - in fact I use Claude through copilot. I have found in general that for the type of work I do (senior to staff level engineering, 90-10 research to programming) that Claude Opus is the o
279.
by devmor 6mo ago
I think you’re handwaving away vague, ungrounded intuition and calling it learning. For instance, if I say “I noticed I run better in my blue shoes than my red shoes” I did not learn anything. If I examine my shoes and notice that my blue s
280.
by devmor 6mo ago
I hate that pattern so much. It’s also not just to obfuscate the spending - it’s also to ensure you already have some amount left over in your account, so that it feels like you’re not spending as much to just “top up” and afford that on
281.
by devmor 6mo ago
You can get better at something without understanding why, but you should be able to think about it and determine why fairly easily. This is something everyone who cares about improving in a skill does regularly - examine their improvement,
282.
by devmor 6mo ago
There’s also just the negative association factor. I use LLMs in my every day work. I’m also a strong critic of LLMs and absolutely loathe the hype cycle around them. I have done some really cool things with copilot and Claude and I keep sh
283.
by devmor 6mo ago
[flagged]
284.
by devmor 6mo ago
The highlighting isn't what matters, its the pretext. E.g. An LLM seeing "```python" before a code block is going to better recall python codeblocks by people that prefixed them that way.
285.
by devmor 6mo ago
This is just persistent memory RAG. I have had a setup like this since about a day after I started using copilot, except it's an MCP server that uses sqlite-vec and has recall endpoints to contextually load the proper data instead of a
286.
by devmor 6mo ago
Why would this be surprising? That’s exactly how much of the code they were trained on is presented in PRs, Forums, etc.
287.
by devmor 6mo ago
Which is very recent. Discord only added full markdown support a couple years ago.
288.
by devmor 6mo ago
Then there's no verification step, preventing the entire mechanism of you not getting spammed.
289.
by devmor 6mo ago
Do you have any reading on this? I find it hard to believe something announced a week ago has been “thoroughly tested”.
290.
by devmor 6mo ago
You want a spiritual successor? We have Ghost. You want anything beyond ghost? Find a way to port the vast market of 100,000+ cheap and free themes and components that are available to enable tech-illiterate, low-budget users to basically b
291.
by devmor 6mo ago
I am sure a lot of people and orgs are - but realistically the majority of users need to understand and prepare not for local-first, but for the fact that they will never have that option for the models they know are the most useful to them
292.
by devmor 6mo ago
Local models just don't seem that useful for me for these particular tasks yet - the most recent versions of Codex and Claude Opus are the first time I've found them to be particularly useful in a "real engineering" cont
293.
by devmor 6mo ago
This is really cool to see! I've been using BM25+sqlite-vec for contextual search projects for a little while, it's a great performance addition.
294.
by devmor 6mo ago
This is a real danger that I think a lot of people will run into as prices go up more and more in the future. Completely outside of the productivity debate, offloading cognitive tasks to LLMs leaves you less practiced in them and less ready
295.
by devmor 6mo ago
That does not work between monitors.
296.
by devmor 6mo ago
I have a lot of complaints but I would say my three big gripes are: - Window navigation within (rather than between) open programs. Mainly if one is on an external monitor, this is just a nightmare and I end up using expose and clicking the
297.
by devmor 6mo ago
I do adore the macOS trackpads, they blow anything else I’ve used out of the water. The keyboard (physically) is also just very pleasant to type on.
298.
by devmor 6mo ago
“Flawless” is absolutely the opposite of how I’d describe the third class keyboard navigation in MacOS. It’s actually more intuitive to use a magic keyboard on the iPad than on the desktop OS.
299.
by devmor 6mo ago
I still don’t think the model is quite great until the bandwidth problem is solved in a way that doesn’t make it prohibitively expensive for alternative appview hosting. It’s the one part of the whole system I think needs a lot of work.
300.
by devmor 6mo ago
If you are looking at it from a business perspective, there is little value to fixing a bug that is not impacting your revenue. Of course, the developers should be determining if the bug may have a greater impact that will or does cause a p

More ›