Your first prompt, step by step. One weak prompt improved version by version (open History to see each change), plus three small everyday prompts that use variables. Copy one, fill the blanks, paste it into Claude, Cursor or any AI tool.
ple import ple:zeynepaslierhan/start-here --allPrompts4
Start here. Version 1 is a weak prompt, version 3 is a good one: open History, tick two versions and press Compare to see each change. Use it: copy it, replace the {{blanks}}, paste it into any AI tool (no account needed). Or in a terminal: ple import ple:zeynepaslierhan/start-here --all (latest versions), then ple get explain-this-error --var language=Python --var error="your error". Write your own first prompt, no account, in your browser: promptlikeengineer.com/try. When: you hit an error you do not understand · Input: {{language}}, {{error}}, {{what_i_tried}} (optional) · Output: meaning, cause, fix, how to confirm
When: you are about to run git commit · Input: {{diff}} (optional, else your agent reads git diff --staged) · Output: a short commit message, subject line plus optional body · In a terminal: ple get write-a-commit-message --var diff="$(git diff --staged)" · keywords: git commit message
When: you wrote a prompt and the answers are mediocre · Input: {{my_prompt}}, {{goal}} (optional) · Output: which of four basics are missing, a better version with variables, one question back · keywords: improve prompt, prompt help
When: you are reading code you did not write · Input: {{level}} (who you are, e.g. new to programming, or senior but new to this codebase), {{code}} · Output: big picture, plain-language walkthrough, the one thing to watch, a tiny experiment · keywords: explain code, learn