The Prompt Pattern Library
About this resource
A pattern is not a prompt. A prompt solves one task; a pattern is a structure you apply to a whole category of tasks for the rest of your working life. Learning eight patterns well produces more capability than collecting two hundred prompts, because patterns transfer to work you have not encountered yet.
This library organizes prompt patterns by the job each one does: assigning a role, supplying worked examples, decomposing a complex request into steps, asking for structured critique, forcing consideration of alternatives, requesting output in a defined shape, iterating on a draft, and having the model interview you before it answers. Each entry gives the structure, a worked example, and the situation it suits.
A prompt solves today. A pattern solves the category. Eight patterns learned properly outperform two hundred prompts collected and forgotten.
How to use it
Learn two patterns properly rather than skimming all of them. The worked-example pattern and the interview pattern between them cover most of the distance from competent to genuinely good, and both feel unnatural the first few times. Apply each to real work three times before moving on, because pattern fluency comes from repetition rather than from reading.
Inside the download
What role assignment actually changes, and when it does nothing.
The single highest-return pattern, and how many examples to include.
Why one instruction with five parts underperforms five sequential ones.
Getting the model to find the weaknesses in its own output.
Common questions
A reusable structure applied to a category of tasks rather than a script for one task. Assigning a role, supplying worked examples, and decomposing a request into steps are all patterns: they work regardless of the subject matter, which is what separates them from a saved prompt.
Including two to five worked examples of the output you want before making your actual request. It is consistently among the most effective adjustments available, because showing the pattern communicates far more precisely than describing it, particularly for tone, structure, and level of detail.
Largely yes, because patterns concern how a request is structured rather than any tool-specific syntax. Specific phrasings and formatting conventions vary between assistants and between model generations, but the underlying structures transfer, which is why they are worth learning over collected prompts.
Free. No email list, no follow-up sequence.
More from the resource library, chosen because they pair well with this one.
Ready-made instructions for getting tables, structured summaries, and clean paste-ready output back.
Customizable prompt templates for customer service, administration, healthcare, retail, finance, and field operations.
What actually makes something an agent, why it changes the accountability question, and what to ask a vendor selling one.
Tell us a little about yourself and this document is yours. We don't have an email list. We collect this so we know who's finding our resources useful, and that's it.
You're downloading
this document
Your information is kept confidential and never shared with anyone.
