The post being replied to essentially said “I don't use the features that have been regularly broken in recent times, or where the features I do use [core git] were broken it luckily didn't affect me, so anyone thinking of leaving has a mental problem”.
Or to paraphrase the old joke:
Q: How many programmers does it take to change a lightbulb?
A: The sunlight through the windows here is working fine, if you can't see where you are that must be a “you” problem.
"Writing software doesn't require GitHub". Well, if they do not need whatever is specific to github and claim that someone who sorely lacks these features has a mental problem...
To be fair, the author of the post said the same thing. From the other thread on HN, they themselves said: "Nobody should cry over a SaaS, of all things. But GitHub has meant so much more to me than that (all laid out in the post). I have an unhealthy relationship with it. "
When a new app is released, it takes a few days for it to get into search, for some reason. Pretty much every single time a new app releases I see a comment like this. Nothing malicious you just have to wait a bit.
The first result is a sponsored result, and even after Friendster is indexed, if they don't pay apple's extortion-rate, the first result will still be for some other social media app.
Ads in the app store are malicious. There are people who have searched "Ledger bitcoin wallet", clicked the first link (a malicious app who paid apple enough money to be 'sponsored' for that search), and had all their money stolen.
nobody has ever claimed that apple makes great cloud software, but all of their walled garden gate-keeping aside, they’re still the last bastion of mainstream local-first computing
Yeah, sure. How do I sync my iCloud photos to my local NAS on linux? A: use a third-party app, they don't build their own.
How do I build an app for my iPhone locally and run it without ever connecting to their servers? I can do that for my phone running linux or for my phone running android, but on iOS I have to get signed by their servers to run code I wrote.
Linux respects my freedom to have my data exist locally, to build and run open source apps, and to modify the code on the devices I run.
Apple does not. They don't let me use their ecosystem from Linux, they do not let me patch the iOS kernel and run a modified version on the devices I run, I can't even access the source code for the macOS kernel.
Apple's filesystem abstraction and lack of something like android "intents" also makes it wildly difficult to do "local-first" computing where files are shared between apps cleanly.
You missed the "mainstream" qualifier from the parent. I am afraid nothing you described here could be considered mainstream, although I'd like to see these things becoming mainstream.
I think it's a warranted attitude to take a failure to implement plaintext search as part of your text search algorithm as enemy action. Yesterday I ended up at Bing for some reason and typed in the unique name of the site I was looking for and it didn't even come up in the first page; All their competitors did, as well as a couple random SEO farms. This is enshittification to the point of unusability, for an application space that we solved in the mid 90's.
My workplace does this to our customers too, where you get worse-than-plaintext-search effectiveness, and I guess it must be profitable enough conning the customer to waste our time as well, as we use the same interface for a lot of customer questions.
This is what killed search more than anything else - people actively working to get their results higher than they would have been under "normal" conditions.
I clicked this thinking “oh, cool, someone finally made a portable version of the Claude.ai* memory system!”
Spoiler, no, it’s not it at all, it’s just a “store”/“remember” memory system… as opposed to the Claude.ai memory system, where it doesn’t make the model actively have to write memories on its own, but rather has a model in the background go through your chat history and generate a summary from it.
I’ve found the latter approach to work much, much better than simple “store”/“remember” systems.
So, it just feels misleading to say this can do what Claude.ai’s can do…
(I’ve been looking for a memory system that works the same for a while, so that I can switch away from Claude.ai to something else like LibreChat, but I just haven’t found any. Might be the only thing keeping me on Claude at this point.)
-
*I say Claude.ai because that’s specifically what has the system; Claude Code doesn’t have this system
I really want to try this approach. I'm curious because this has not been my experience at all. I created https://github.com/flippyhead/ai-brain mostly just for myself and a few friends use it. But so far, telling the AI (via CLAUDE.md) to look for relevant memories and to think about when and how to save them has worked very well. It can create structures based on decided priorities, notes for the future, that feel like they'd be very different if it was just trying to summarize everything.
I use Claude code hooks to prompt and store memories. It’s taken a lot of iterations mostly on the definition of “significant” events being stored in memory. Indeed, it works very well now but I’m hesitant to start from scratch on some guys tool.
I think demos are going to need reviews here on out. Vibe coded projects look too legit but it’s a waste of time to test the 100 that come out each day
I hear you. I've been slowly building up my own tool (linked above) and keep feeling like someone is going to soon release something that a lot of people will agree should be an independent standard. I'm reluctant to host it with someone else so it needs to be opensource. But then again what I've got is working well for me.
The biggest issue for me is recalling during conversation context, not jotting information down. I've solved this by including a tag for when to nudge the agent to recall something.
Shameless drop: My own agentic environment is also using a summarizer to sum up agent histories when they overflow in their context windows. Additionally, all short lived agents are based on requirements (WIP) as a centralized point for code, unit tests, and architecture.
(Everything is tailored to Go as a language)
Works pretty good so far, the user only interacts with the planner. I'm working atm on the requirements to have a spec driven workflow. Web UI is the most polished atm because of ability to have agent tabs on the side for better overview.
I favor automatic recall, invisible to the agent. For memory creation, I find tool calls do a pretty good job, though I also like automatic memory creation on context compression.
I think with automatic creation you need async consolidation (calling it dreaming is a little dramatic for my taste).
That's an interesting concept.
So it's like if you're an agent chatting with a user, you have an army of assistants who overhear the conversation and record important facts, or search relevant facts on some database and decide on the fly when to interrupt you with "this memory X looks relevant". Sounds easy enough if tokens were free, but an interesting problem to do it efficiently.
Burst-parallel non-frontier models can resemble "tokens were free". And there one might potentially augment not just conversations, but CoT - retroactively by submitting messages with altered reasoning strings, or inline with the inference loop watching CoT and attempting non-distracting injection.
Simple vector similarity plus a cheap model to filter results works pretty well. Though ofc t does add tokens to your primary chat, which is the basic tradeoff of memory systems in general (in addition to latency)
That's exactly what claude-code does these days. If you AFK for ~5 minutes it also produces a summary of where you are, which is useful if you're juggling multiple windows.
How do you benchmark that? The problem with extracting memories in the background is it's hard to make that work with the prefix cache. You can go a long way with a simple 2-stage LOG.md (detailed log of tasks & lessons) + MEMORY.md (log items that are promoted when the log itself gets truncated) + a stop hook to ensure this runs at the end of a turn.
Yeah. The other advantage is a summary-based memory also just… “pieces together” things that a “store”/“remember” memory wouldn’t, because they’re things that the actual main agent would not think to store. i.e. small disconnected things across conversations that alone, would not end up in memory because they’re insignificant. But when there’s an agent looking at multiple conversations at once it can actually reason about this stuff and piece it together.
Just write a hook that runs claude -p after whatever you want and update whatever memory system you want. You can use a channel to inject back what topics were update or what have you.
Sure, but the idea is not to have this in Claude Code, the idea is to be able to use something like LibreChat with proper memory. I don’t really need that good of a memory system for my coding agent, it’s definitely more something I need for my chat agent.
Which is weird because if I remember correctly, there are summaries already generated by Claude on your hard drive of things you have done in the past.
Termius integrates TPM/Secure Enclave keys nicely on Windows Mac iOS etc as well, which is nice because then you can have a key on your phone (so if you lose your laptop you won’t get locked out of your servers)
What are you using for virtual files? I definitely found it to work much much better when I added Nextcloud through Gnome’s settings rather than the Nextcloud client
You mean via WebDav? That is definitly an option, but then it is not possible anymore to mix working file directories within Nextcloud. E.g. I have a long list of ignored files for nextcloud [1], such as .git folder etc.
I would need to create a separate file tree, splitting my work between nextcloud maintained files and working files. That feels cumbersome. On Windows, I could directly add my whole drive to nextcloud as a virtual file system.
