Prompting prompts

Better outputs from any AI

15 prompts · Build

• Add external knowledge to any prompt without fine-tuningRetrieval-augmented prompt design that structures retrieved content for maximum utility — covering selection, formatting, and the prompts that use it well.
• Add structured reasoning to any prompt to get dramatically better outputChain-of-thought techniques that force the model to show its work — producing more accurate, more auditable, and more trustworthy responses.
• Add the guardrails that stop an AI from going off-scriptConstraint injection techniques that limit topic scope, output length, tone, and behaviour — so the model does the job without the surprises.
• Break your own prompt before someone else doesAdversarial prompt testing that checks for jailbreaks, edge cases, and the inputs that produce outputs you didn't expect — before you ship it.
• Build the prompt library your whole team can use and improveA prompt documentation and versioning system with the metadata, change history, and testing protocol that turns a collection of prompts into a shared, maintainable asset.
• Build the system prompt that shapes how an AI thinks and respondsA system prompt architecture for agents, assistants, and bots — covering persona, constraints, output format, and the guardrails that keep behaviour consistent.
• Chain prompts together so the output of one feeds the nextMulti-step pipeline design that breaks complex tasks into stages, passes outputs between steps, and handles the edge cases that break naive chaining approaches.
• Dial in the creativity vs. precision tradeoff for your specific taskTemperature and parameter guidance for common task types — when to go low for factual accuracy, when to go high for creative range, and everything in between.
• Find exactly why a prompt isn't working — and fix itA systematic prompt debug process that identifies whether the failure is in the instruction, the context, the format, or the model's limitations — and fixes the right thing.
• Get clean, parseable output from any model, every timeA structured output design process covering format specification, delimiters, schema definition, and the prompting techniques that keep complex outputs intact.
• Get more signal into the context window without hitting the limitContext window optimisation strategies for long documents, complex tasks, and multi-turn conversations — including compression, chunking, and retrieval design.
• Give an AI an identity that makes every response more usefulPersona and role design for AI assistants — the identity, expertise, and constraints that shape every interaction toward the outcomes you actually need.
• Grade AI responses with criteria so you know what 'good' looks likeAn evaluation rubric for AI output that defines quality explicitly — so you can compare responses, improve prompts, and stop judging by feel.
• Teach a model to behave exactly how you want — with examplesFew-shot example design that communicates the pattern you want through demonstration — the fastest way to get consistent behaviour without fine-tuning.
• Use AI to write better prompts than you'd write yourselfA meta-prompt that takes your task description and produces a high-quality prompt — with the context, constraints, and output spec you'd have taken an hour to write.

All prompt categories · Prompt libraries you can download