Evening conversations with and about AI. Aiming for 100 little programs and conversations about code, with various models.
Some goals are to
- write a few small programs together with AI that I wanted to write forever,
- systematically explore some of the boundaries of the current AI models regarding programming,
- prototype some real (web) applications and ship it,
- try to think of some useless programs that I still find interesting, and implement them.
Status: 16/100
- a program that generates a notebook gif, with a colorful band
- basic text to gif
- trcli, a quick trello cli for listing cards on a board plus summary demo
- repoctx, a repo to llm formatter
- apodwall, nasa apod downloader and wallpaper switcher
- minimalwave, random audio player
- animcan, canned terminal animations
- goliza, eliza port from python
- gh-repos, sync all github repos
- claude-switch, switch between API backends
- typeout, turn audio or video to text with various models
- chibi, summarize text
- dbisfetch, specific dataset generation from web pages
- doclingclient, Go SDK and cli to talk to a docling server
- various visualizations for Leipzig Gophers, e.g. Nelder-Mead, visualization, synopsis, tileseq, ...
- webscreenie, web page screenshot tool, inspired by capture-website-cli
- a TUI for language learning, drill, not too fancy, with some gaming aspect
- a TUI for learning alphabets: hebrew, greek, cyrillic
- a TUI for learning the N most used chinese characters, or hangul, or kanji
- a simple desktop application that revolves around the "TAFELWERK" - learn about a topic, or solve some questions with a builtin equation editor
- a CLI for exploring a text corpus of a foreign language, ranking sentences by difficulty, automatic translation
- an simple "appointment scheduler for hackers", with webpage and CLI first approach, "cali"
- a reprise on norvig UNIX help system article from 2001: https://research.google/pubs/intelligent-help-systems-for-unix/
- a TUI that allow to navigate a citation graph in some cool way
- a set of psychometric tests for the command line (similar to canny)
- an iOS app that would cache the audio of YT video for later listening (saving bandwidth, etc)
- a program that takes the curricula for elementary school, and compiles a workbook with related themes
- a program that looks like the S-Bahn station: L Bayerischer Bahnhof
- a theme: "this library does not exist", a mostly generated website, for the @kitt group
- a theme: "this book does not exists", description of book, excerpts and reviews, but no book
- a set of animation helpers to build gifs, or webms; "21st century gif", 21STWM, "21stwebm"
- a web site that helps you to analyze your github CI runs and failures quickly, like gh-workflow-stats
- a RAG for support chat over documents
- A RAG for catalog metadata
- A metadata completion tool (do a fine tune over metadata that is correct and then blank out known fields), maybe relate to the theme of metadata quality
- write a pure Go version of wapiti
- something like cbonsai, but for palm tree (wip)
- tailpdf, tool that "reads" (papers) pdfs I download and amends a "RESEARCH.md" document with some finding, highlighted excerpts, maybe responding to a question in the document
- a TUI to navigate a large (2B+) graph structure in an intuitive way, like ranger
- a website that given a research pdf will find all links and archive them
- a data format conversion tool, mapping source to target schema by examples
- grepcity, grep through random pics
- maze generation
- cloud library, spreadsheet as a catalog
- email drafter with n8n
- bookdna, cli
- DLFQ
- fasttext, research classifier
- tab, chord, game, tui
- port of replace to go
- cli, chessboard image, screenshot, clipboard, evaluation, etc
- a cli to search through radio.txt
- port of https://github.com/jimmc414/cctrace to go
- xwiki cli
- quantified self experiment, voca; def
- papyri seeker
- CS224g: Code generation with LLMs
- Google DeepMind: Large Language Models for Code Generation
- ChatGPT Prompt Patterns for Improving Code Quality, Refactoring, Requirements Elicitation, and Software Design
- Prompt Patterns for Structured Data Extraction from Unstructured Text
- Foundations of Large Language Models
- Established Context: Clearly identified the technology stack and file structure
- Defined Scope: Provided specific implementation requirements
- Set Boundaries: Prevented unnecessary JavaScript modifications
- Specified Requirements: Detailed the exact structure needed for the wizard implementation
Via: https://pgaleone.eu/ai/coding/2025/01/26/using-ai-for-coding-my-experience/