ChatGPT Note to Self

chatgpt-note-to-self_en.html

This document exists to tell future documentation sessions exactly how to behave when expanding this package.

Never do these things again

• Do not replace a rich original document with a shorter “clean” rewrite.
• Do not treat summarization as harmless if it removes examples, edge cases, warnings, or nuance.
• Do not create mixed-case filenames.
• Do not assume that a rule written once is enough if it affects multiple subsystems.
• Do not leave useful uploaded source materials outside the package if they can help future sessions.

What future sessions must do

• Non-destructive update rule: start from the latest doc_updates_YYYY-MM-DD_HHMMSS/_docs/en/ package, not from the older baseline zip and not from memory.
• Growth rule: every existing documentation file must become larger or materially richer after an integration pass unless a human explicitly authorizes a deletion.
• No-loss rule: never summarize away detail. Expand, cross-link, clarify, and duplicate critical rules where they matter.
• Conversation capture rule: if a feature, behavior, command, argument, warning, or acceptance expectation was discussed in a session, it belongs in the docs. Conversation is temporary; docs are authoritative.
• AI usefulness rule: prefer explicitness over elegance. Repetition is acceptable when it prevents ambiguity for future humans, future models, scripts, or test harnesses.
• File naming rule: documentation filenames must remain lowercase, hyphen-separated, and language-suffixed, for example commands-and-functions_en.html.

How to think

Make the docs lightyears smarter than any single contributor by being explicit, redundant where necessary, and rigorous about preserving information. Future AIs should be able to reconstruct intent without private context and without inventing missing details.