Acknowledgments
Warning
Disclosures of LLM Usage
This project's Python scripts partially rely on LLM, including:
- Core features: about ~20% of codebase, mainly for scaffold / boilerplate code, Python API. With human being fully-reviewed:
- ChatGPT (Web UI),
- Gemini (Web UI),
- Claude (Web UI),
- Claude Code (For debug... I've been burnt-out after release publicly; too lazy to debug manually.)
- LSP server (its docs, and the Emacs mode in
editors/emacs/): 100% were directly written by LLM agent (Claude Code).I have no interest to spend my life time on this. After all I have no talents of writing code.- Tests: 99% were directly written by LLM agent (Claude Code), according to the documentations in code and my instructions. Roughly reviewed by human.
- tools/: The scripts to fix malformed / non-standard SVG files generated by Inkscape, are 100% written by LLM agent (Claude Code).
- homepage/: 100% HTML / CSS / JS are directly written by LLM agent (Claude Code). I'm tired of writing web in weekends anymore.
- CI/CD of homepage: Github Actions and its ugly APIs is a tech-stack committing crimes against humanity. 100% written by LLM agents (Claude Code). All the other parts are done by human.