Give the model a job, not a wish
A role narrows vocabulary and judgement. 'You are a senior editor' beats 'help me write' because it implies standards, not just output.
Read the beginner guideSearch prompts, categories, platforms and repositories
Most prompting advice is a list of tricks. This hub is a sequence: six fundamentals that account for most of the quality difference, followed by in-depth guides on specific workflows. Start at the top, apply each idea to one real task, and only then move on. The library has 520 worked examples when you want to see the pattern in practice.
A role narrows vocabulary and judgement. 'You are a senior editor' beats 'help me write' because it implies standards, not just output.
Read the beginner guideThe same facts need different framing for an executive, a customer and a developer. Audience is the cheapest quality improvement in prompting.
Browse writing promptsSpecify format, length and structure before generation. You spend less time reformatting and more time judging the substance.
Build a promptWords to avoid, claims to verify, sections to include. Constraints are where domain knowledge enters the prompt.
Diagnose a promptOne short example of the voice you want outperforms five adjectives. Models imitate patterns more reliably than they follow style labels.
See a worked exampleAsk the model to critique its own draft against the brief, then rewrite. Two passes usually beat one long prompt.
Find a revision prompt10 articles across 6 topics.
This week: a system prompt for research agents, a safer browser automation pattern, and three repositories gaining stars fast.
The frameworks, inference engines, agent tools and learning resources that are actually worth your time, with licence notes and what each is best for.
Marketing, customer service, pricing and admin prompts for teams without a dedicated AI department, plus the guardrails to keep you out of trouble.
Most 'study prompts' generate summaries that feel productive and teach you nothing. These twelve prompts are built around retrieval, feedback and self-testing.
The same prompt rarely produces the same result on both models. We tested structure, tone, long documents and coding tasks to find where each one has an edge.
Generative engine optimisation is not a trick. It is clear structure, original evidence, consistent entities and pages that are easy to quote.
A repeatable structure for Midjourney v7: subject, light, lens, composition and parameters. Plus the mistakes that make images look generic.
A five-block structure for system prompts, with examples, the rules that cannot be enforced in a prompt, and how to test for drift.
The techniques that matter, when to use them, and why chain-of-thought is not a universal upgrade. With copy-ready patterns.
What retrieval-augmented generation actually does, where it fails, and the questions to ask before you promise a 'chat with our docs' feature.
Every prompt in the library includes a worked example, tips and failure modes.