But now kids grow up with social media and fake news. I too grew up seeing stuff I shouldn't have online and I turned out fine but I see the online landscape today as much scarier place and as a recent father I'm scared of what it can do to my child and will need to police it.
I'm not entirely sure what an agentic workflow could mean today but I think so. I use a coding agent (crush), prompt it to brainstorm an implementation with me (or sometimes I know exactly how I want to implement it but ask it to challenge it), correct any wrong assumptions or request the implementation to look differently than suggested if I don't like it. Then finally when I'm positive I've cleared the most important assumptions I ask it to actually write and edit files and run tests and such (this just ends up being a "implement this").
With any model I've tried I've found it to be a huge pain to have it fix things where it made a wrong assumption without the code becoming a mess and burning a lot of tokens. I'm aware that not everyone works like this but I'm still very opinionated on what the end result should look like so I can still work on it without an LLM.
How are you using it? I have the lite plan and I've only ever maxed my weekly usage a few hours before reset. I will concede that I'm not a super heavy LLM user but it's been really good for me.
My workflow is usually:
- read file. I want to achieve X, how do? Do not implement anything.
- I would do a, b and c
- sketch a brief implementation of your suggestion
- <code> (not writing files yet)
- instead of your approach x, wouldn't it make sense to instead do z? What would that look like?
I remember that there were these talks called "Linux sucks" (IIRC) and the premise wasn't that we should start using Windows but more like "we can do better and improve it." Not every complaint needs to come with a solution or better alternative.
I made an MCP server that basically implements all the tools an agent harness would provide. The code is checked out on my server, the MCP server creates git worktrees on "activation" and it can read, modify files, run bash commands etc. I have this setup in typingmind so I can do everything from my phone :p The only problem is that typingmind needs to be in the foreground on my phone, otherwise it will kill the connection.
There was a time where I was often stuck for an hour with nothing but my phone and I kept copying file contents into chat for context so I made this and it works surprisingly well.
Generally I think you have a point when making landing pages or things of that sort. I however basically only use static site generators for blogging and I think markdown/asciidoc/whatever + metadata is nicer for writing content than directly in HTML.
Yeah it would probably be more efficient to do the templating as a build step in the derivation that parses the markdown. But after making nixidy[1] I wanted to have a crack at making a static site generator _in nix_. I'm open to expanding it though of course :)
I had seen a few of those `__findFile` hackery articles lately and did actually consider doing that but I think having a nice way of adding HTML attributes would've been clunky :p
Edit: although looking at this article it seems to be supported.
I'm the author of nixtml. It's already doing IFD by using python to parse markdown content + metadata and superhtml to format the final HTML (a bit unnecessary, but I liked it while developing it).
I really wanted the templates to just be nix functions. It shouldn't be an issue to pass the context to an external program with `pkgs.runCommand` or something and then read the result (IFD like you mentioned).
No one claims their showels find gold for you.