Nimo π³οΈβπ
@nimobeeren.com
he/him
Building cool things with or without AI (mostly with) β π§ͺπΉπ»πΈπ·π§π
π nimobeeren.com
π Eindhoven
Building cool things with or without AI (mostly with) β π§ͺπΉπ»πΈπ·π§π
π nimobeeren.com
π Eindhoven
Pinned
GitHub - nimobeeren/strands-solver: A solver for Strands, the New York Times puzzle game.
A solver for Strands, the New York Times puzzle game. - nimobeeren/strands-solver
github.com
Finally published strands-solver 1.0.0 π
It solves NY Times Strands puzzles using a backtracking algorithm and word embeddings.
Try it with `uvx strands-solver solve today` (free tier GEMINI_API_KEY optional)
github.com/nimobeeren/s...
It solves NY Times Strands puzzles using a backtracking algorithm and word embeddings.
Try it with `uvx strands-solver solve today` (free tier GEMINI_API_KEY optional)
github.com/nimobeeren/s...
Iβve been getting so many chores done while Claude is working π§½
I try to avoid context switching, even if that means Iβll be less βproductiveβ for a bit while my agent works. Iβll pick up a household chore while daydreaming about the project, perhaps thinking about what to build next. Definitely takes conscious effort not to go full frenzy mode though.
February 9, 2026 at 6:37 PM
Iβve been getting so many chores done while Claude is working π§½
I try to avoid context switching, even if that means Iβll be less βproductiveβ for a bit while my agent works. Iβll pick up a household chore while daydreaming about the project, perhaps thinking about what to build next. Definitely takes conscious effort not to go full frenzy mode though.
February 9, 2026 at 6:37 PM
I try to avoid context switching, even if that means Iβll be less βproductiveβ for a bit while my agent works. Iβll pick up a household chore while daydreaming about the project, perhaps thinking about what to build next. Definitely takes conscious effort not to go full frenzy mode though.
Iβve definitely been feeling this too. I wonder to what degree this can be solved by just making the tools faster. It feels like context switching is the real killer and having to wait for 10-15 mins after each prompt just makes that very likely to happen.
February 9, 2026 at 6:37 PM
Iβve definitely been feeling this too. I wonder to what degree this can be solved by just making the tools faster. It feels like context switching is the real killer and having to wait for 10-15 mins after each prompt just makes that very likely to happen.
Agentic coding is starting to feel cost-bound again for me. I use API-based billing (bc work) and Iβm no longer sure that what I build is necessarily worth the price tag.
$20 for a feature is not much, but also not negligible. And I donβt like that I have to consider that before starting.
$20 for a feature is not much, but also not negligible. And I donβt like that I have to consider that before starting.
February 9, 2026 at 6:21 PM
Agentic coding is starting to feel cost-bound again for me. I use API-based billing (bc work) and Iβm no longer sure that what I build is necessarily worth the price tag.
$20 for a feature is not much, but also not negligible. And I donβt like that I have to consider that before starting.
$20 for a feature is not much, but also not negligible. And I donβt like that I have to consider that before starting.
Iβm not sure constraints are essential for open-endedness per se, but they do breed creativity.
February 8, 2026 at 2:53 PM
Iβm not sure constraints are essential for open-endedness per se, but they do breed creativity.
Hmm, thatβs a good distinction. Terminals are definitely open-ended in the sense that you can end up in a wild variety of places, as long as they are text-based.
February 8, 2026 at 2:53 PM
Hmm, thatβs a good distinction. Terminals are definitely open-ended in the sense that you can end up in a wild variety of places, as long as they are text-based.
I think CLIs are winning precisely because they are *not* very open-ended. Youβre fully constraining yourself to text input/output, and thatβs great for LLMs.
Why is the AI revolution happening on the command line first? Because the terminal is still the most open-ended general-purpose computing platform available.
Unix Philosophy is All You Need.
Unix Philosophy is All You Need.
there has never been a better time to get terminal-pilled
February 8, 2026 at 12:55 PM
I think CLIs are winning precisely because they are *not* very open-ended. Youβre fully constraining yourself to text input/output, and thatβs great for LLMs.
I hope that LLMs will help us make that transition. The hardest part in these large scale migrations (for me) is understanding the exact behavior of the current system. I think an AI that independently explores (by trial-and-error) and documents systems would be super useful.
February 8, 2026 at 12:48 PM
I hope that LLMs will help us make that transition. The hardest part in these large scale migrations (for me) is understanding the exact behavior of the current system. I think an AI that independently explores (by trial-and-error) and documents systems would be super useful.
Made a self-updating skill for scaffolding React projects. It runs a script so I get a fast and consistent setup every time, but uses the LLM to update the script based on shadcn docs.
github.com/nimobeeren/s...
github.com/nimobeeren/s...
github.com
February 7, 2026 at 6:32 PM
Made a self-updating skill for scaffolding React projects. It runs a script so I get a fast and consistent setup every time, but uses the LLM to update the script based on shadcn docs.
github.com/nimobeeren/s...
github.com/nimobeeren/s...
Yeah, I was more relieved reading about what it _couldnβt_ do yet than I was excited with what it could
February 6, 2026 at 6:21 AM
Yeah, I was more relieved reading about what it _couldnβt_ do yet than I was excited with what it could
How good is Claude Code at Factorio
February 5, 2026 at 8:32 PM
How good is Claude Code at Factorio
I've been having way more (good) interactions on here for the last few weeks! I was added to a starter pack but I feel like it can't just be that. Probably For You has something to do with it? π
February 5, 2026 at 5:11 PM
I've been having way more (good) interactions on here for the last few weeks! I was added to a starter pack but I feel like it can't just be that. Probably For You has something to do with it? π
I think they'll get better taste. They're excellent at doing what you ask if you're unambiguous. But they'll get better at choosing the right thing to do if you're not.
February 5, 2026 at 5:09 PM
I think they'll get better taste. They're excellent at doing what you ask if you're unambiguous. But they'll get better at choosing the right thing to do if you're not.
Trying to explain nondeterminism in LLMs: it's like talking to clones of the same person but they're in a slightly different mood each conversation
February 5, 2026 at 5:08 PM
Trying to explain nondeterminism in LLMs: it's like talking to clones of the same person but they're in a slightly different mood each conversation
Yes, LLMs can help you be less ambiguous! And to be fair, those optimizers do sometimes make use of some tricks to squeeze last few percentage points of performance out (at least so they claim)
February 5, 2026 at 11:42 AM
Yes, LLMs can help you be less ambiguous! And to be fair, those optimizers do sometimes make use of some tricks to squeeze last few percentage points of performance out (at least so they claim)
Thatβs true, and I guess models are getting better at choosing the right interpretation when there are multiple. Itβs just going to be difficult when you rely on the same interpretation being chosen every single time.
February 5, 2026 at 11:39 AM
Thatβs true, and I guess models are getting better at choosing the right interpretation when there are multiple. Itβs just going to be difficult when you rely on the same interpretation being chosen every single time.
Rambling into a recording device like itβs the 70s, except Iβm not causing psychic damage to the person who is asked to transcribe it
February 5, 2026 at 6:38 AM
Rambling into a recording device like itβs the 70s, except Iβm not causing psychic damage to the person who is asked to transcribe it
The structure is not even strictly necessary for the LLMs, itβs mainly for humans to keep things interpretable and maintainable. You could just word vomit as long as there are no contradictions.
February 4, 2026 at 10:20 PM
The structure is not even strictly necessary for the LLMs, itβs mainly for humans to keep things interpretable and maintainable. You could just word vomit as long as there are no contradictions.
The whole βprompt engineeringβ thing still has people believing you need to speak some arcane language to make good use of LLMs. In reality, you just need to be unambiguous, structured and complete. No tricks necessary.
February 4, 2026 at 10:20 PM
The whole βprompt engineeringβ thing still has people believing you need to speak some arcane language to make good use of LLMs. In reality, you just need to be unambiguous, structured and complete. No tricks necessary.
I think local and open data will become more and more important, with local-first and AT proto setting good examples
February 4, 2026 at 7:06 PM
I think local and open data will become more and more important, with local-first and AT proto setting good examples
If it works, yes π
February 3, 2026 at 8:41 PM
If it works, yes π
Sonnet is definitely due for an upgrade! Opus is actually faster for me in most cases because it calls the right tools instead of messing around.
February 3, 2026 at 7:56 PM
Sonnet is definitely due for an upgrade! Opus is actually faster for me in most cases because it calls the right tools instead of messing around.
Yeah, like please invest that effort into turning your app state into local Markdown and JSON (or SQLite) files so I can use my good AI tools to work with it
i wish all these random programs would stop adding AI assistants. not because AI is bad, but because these programs are bad and i don't want them implementing ai. like adobe acrobat. come on.
February 3, 2026 at 5:09 PM
Yeah, like please invest that effort into turning your app state into local Markdown and JSON (or SQLite) files so I can use my good AI tools to work with it
Really cool! Looks like thereβs a lot of potential for skills like this to speed up agentic work, independent of the models themselves.
February 3, 2026 at 7:25 AM
Really cool! Looks like thereβs a lot of potential for skills like this to speed up agentic work, independent of the models themselves.