Skip to content

"lmao how is this not solved yet?" list #1

@veilm

Description

@veilm

1745723680 01

"find orangeblugt on bandcamp and write a scraper to get a list of their songs and albums. integrate it into my existing RSS scripts"

the LLM needs to be able to do basic visual or text-based browser navigation. it needs to write parsers for HTML, ideally test them itself. it needs to navigate your filesystem, find where you keep your shell scripts, how your RSS system works, and what changes need to be made. you should ideally be able to inspect its history of outputs in a simple way (e.g. just filter by the history of shell commands used, and then the diff)

regular backups and basic rm wrappers (like confirming you want to entirely delete $HOME) > having to babysit the LLM on every move

you don't reincarnate each time you need a new feature so ideally the LLM wouldn't either --> automatically maintain an LLM-friendly memory store. of what? everything:

  • your preferences in music
  • the layout of your filesystem
  • the conventions of your existing parsers
  • the design of your different shell scripts you have installed
  • its own common mistakes in editing
  • its short-term and medium-term plan for navigating the web or implementing the change or whatever it's doing

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions