Hand-written prompts for AWS cert prep, IT CV writing, YouTube shorts, recruiter sourcing. Each one is dated by the model it was tested on and the month it was last verified. No "10 ChatGPT prompts that will change your life" garbage — just things that actually work.
A prompt is battle-tested when it produces useful output across multiple models, multiple times, on multiple variations of the input. We don't ship prompts that worked once on one model — we ship the ones we keep coming back to.
Why so few prompts per workflow?
Quality > quantity. 4-6 prompts that solve real problems beat 50 generic "act as a professional X" templates. If you can name the problem in one sentence, the prompt is worth shipping; if not, it isn't.
Model-tested dates
Every prompt lists the models + dates it was verified against. LLMs change quarterly — re-test any prompt older than 6 months before depending on it.
CertQuests