CodeMonument
@codemonument.bsky.social
Vibe coded PRs suck.
And even LLM assisted PRs might be bad, because you skip thinking about the structure of your code. I think, we should have the respect for the time and effort of open source maintainers to deliver thoroughly researched and coded PRs
And even LLM assisted PRs might be bad, because you skip thinking about the structure of your code. I think, we should have the respect for the time and effort of open source maintainers to deliver thoroughly researched and coded PRs
November 10, 2025 at 12:07 PM
Vibe coded PRs suck.
And even LLM assisted PRs might be bad, because you skip thinking about the structure of your code. I think, we should have the respect for the time and effort of open source maintainers to deliver thoroughly researched and coded PRs
And even LLM assisted PRs might be bad, because you skip thinking about the structure of your code. I think, we should have the respect for the time and effort of open source maintainers to deliver thoroughly researched and coded PRs
around 30 to 60 percent.
BUT: This works fine for me IN MY PERSONAL TOY PROJECTs!
I'm very cautions in my own projects to not introduce slop I cannot control if the llm fails.
Also, one reason for LLMs performing great are foundational tools like vite which encode correct behavior.
BUT: This works fine for me IN MY PERSONAL TOY PROJECTs!
I'm very cautions in my own projects to not introduce slop I cannot control if the llm fails.
Also, one reason for LLMs performing great are foundational tools like vite which encode correct behavior.
November 10, 2025 at 12:07 PM
around 30 to 60 percent.
BUT: This works fine for me IN MY PERSONAL TOY PROJECTs!
I'm very cautions in my own projects to not introduce slop I cannot control if the llm fails.
Also, one reason for LLMs performing great are foundational tools like vite which encode correct behavior.
BUT: This works fine for me IN MY PERSONAL TOY PROJECTs!
I'm very cautions in my own projects to not introduce slop I cannot control if the llm fails.
Also, one reason for LLMs performing great are foundational tools like vite which encode correct behavior.
I got into vibe engineering lately, telling the AI my whishlist of features and what my goals are. After that I'll let it generate comprehensive architecture/ decision documents and tell it to ask me questions about unclear requirements. And I review the things it generates on a low to mid level ...
November 10, 2025 at 12:07 PM
I got into vibe engineering lately, telling the AI my whishlist of features and what my goals are. After that I'll let it generate comprehensive architecture/ decision documents and tell it to ask me questions about unclear requirements. And I review the things it generates on a low to mid level ...
I have an absolute tiny shimmer of what that might mean. But the details are absolutely unclear. Maybe I'll ask ai tomorrow what your post + the code could mean 😃😆
November 7, 2025 at 11:13 PM
I have an absolute tiny shimmer of what that might mean. But the details are absolutely unclear. Maybe I'll ask ai tomorrow what your post + the code could mean 😃😆
Huuuuge! I'm currently not doing anything with SSE events, but I had many times in the last 5 years when I wanted this! 😍 it starts to be a real alternative to websockets!
November 5, 2025 at 11:50 PM
Huuuuge! I'm currently not doing anything with SSE events, but I had many times in the last 5 years when I wanted this! 😍 it starts to be a real alternative to websockets!
I tried it once, it does this crypto stuff and feels overloaded to me. For some reason they don't feel trustworthy to me.
November 2, 2025 at 4:48 PM
I tried it once, it does this crypto stuff and feels overloaded to me. For some reason they don't feel trustworthy to me.
Sounds really good! I probably have to let it be explained to me by ai at the beginning 😆. But this sound likes very good starting/upgrade path for my own beginner nix-Darwin flake which manages a tiny bit of my macOS apps 😃😂
November 2, 2025 at 5:52 AM
Sounds really good! I probably have to let it be explained to me by ai at the beginning 😆. But this sound likes very good starting/upgrade path for my own beginner nix-Darwin flake which manages a tiny bit of my macOS apps 😃😂
Reposted by CodeMonument
- IndexedDB to store non-serializable values (like the currently open FileSystemFileHandle).
- Web Workers for binary manipulation and expensive compute.
- Popover API for accessible hint/popover-like elements.
- <dialog> for native, accessible, and stylable modals and dialogs.
- Web Workers for binary manipulation and expensive compute.
- Popover API for accessible hint/popover-like elements.
- <dialog> for native, accessible, and stylable modals and dialogs.
October 31, 2025 at 1:39 PM
- IndexedDB to store non-serializable values (like the currently open FileSystemFileHandle).
- Web Workers for binary manipulation and expensive compute.
- Popover API for accessible hint/popover-like elements.
- <dialog> for native, accessible, and stylable modals and dialogs.
- Web Workers for binary manipulation and expensive compute.
- Popover API for accessible hint/popover-like elements.
- <dialog> for native, accessible, and stylable modals and dialogs.
Yassss, nice wordplay there, appreciate it! 😁
October 31, 2025 at 6:39 AM
Yassss, nice wordplay there, appreciate it! 😁
Thanks! 😃 I cannot try it, since I'm not working in that project anymore. But I'll try it again for the many Webapps I have personally. They have way less custom config
October 31, 2025 at 6:30 AM
Thanks! 😃 I cannot try it, since I'm not working in that project anymore. But I'll try it again for the many Webapps I have personally. They have way less custom config