Build Your Own Prompt Library
Stop rewriting the same prompt. Build a small, organised, tested personal library — and learn the versioning habit that stops it rotting into a folder of near-duplicates.
Why this one
Almost everyone has prompts scattered across chat history, notes and screenshots. The cost is not storage, it is that you rewrite a worse version of a prompt you already perfected two months ago.
Steps
Mine your history for what you actually reuse
Go through your last month of AI conversations and pull out every prompt you have written more than once, in any form. That list is your real library. It is usually shorter and more boring than the one you would design from scratch, which is the point.
Give each one a fixed record
For every prompt, write down: what it is for, the prompt itself, an example of good output, and what it is bad at. The last field is the one everybody skips and the one that saves you from using a prompt in a situation it was never good at.
Organise by task, not by tool
Group by what you are trying to do — summarising, drafting, critiquing, extracting — rather than by which model you used. Tools change every few months; tasks do not, and a library organised by tool needs reorganising every time you switch.
Test each prompt on a fresh input
Run every prompt once against input it has not seen. Some will fail, because they were quietly relying on context earlier in an old conversation. Fix or delete them now rather than discovering it under time pressure.
Version instead of overwriting
When you improve a prompt, keep the old one and note what changed and why. You will occasionally find the "improvement" was worse for a case you had forgotten about, and without the old version you cannot go back.
You should end up with
An organised, tested personal prompt library with usage notes.
Done when
- Every prompt has a documented weakness, not just a purpose
- Each one has been tested on input it has not seen before
- You can find the right prompt in under thirty seconds
If you want to go further
- Add a "prompts that failed and why" section — it is the most instructive part
- Share three with a colleague and note where they had to change them
Useful alongside this
Try next
Topics
Related across the site
Reasoning model
A reasoning model is trained to generate extended internal deliberation before answering. It trades latency and cost for substantially better performance on maths, logic, coding and multi-step problems.
Prompt caching
Prompt caching stores the processed form of a repeated prompt prefix so subsequent requests reusing it are cheaper and faster. Cached input tokens typically cost a fraction of normal input tokens.
AIDA and PAS Prompts — ChatGPT
AIDA (Attention, Interest, Desire, Action) and PAS (Problem, Agitate, Solution) are proven copy structures.
Fine-tuning
Fine-tuning continues training a pre-trained model on your own examples to specialise its behaviour. It is the right tool for teaching consistent style, format or classification behaviour — not for adding knowledge.
GPT vs Claude vs Gemini: An Honest Comparison
A practical comparison of the three leading AI models across writing, coding, reasoning, long documents and cost — including which one to reach for on which task, and where each genuinely falls short.
Meta-Prompting: Prompts That Write Prompts
The fastest way to a great prompt is asking the model to write and critique one.