Skip to content
PromptifyLab

Prompt collection

Prompt Engineering

56 prompts Free · no sign-up Works in ChatGPT, Claude & Gemini Search & filter these

Fundamentals, techniques, system prompts, chaining, model guides, structured output, troubleshooting and safety. 56 copy-ready prompts across 8 categories – fill in the brackets, copy, and paste into ChatGPT, Claude, Gemini or any capable assistant.

This pillar is the method behind the other eight. It is deliberately small on theory and long on things you can test — because most prompt-engineering advice is folklore that was true for one model on one day and got repeated until it sounded like a rule.

Where the evidence is thin, we say so.

8 categories

  • Fundamentals

    The handful of things that account for most of the difference.

    7 prompts

  • Techniques

    Few-shot, chain of thought, role prompting — what holds up and what has been superseded.

    7 prompts

  • System Prompts

    The instruction that ships with every message, written so it is still followed at message forty.

    7 prompts

  • Chaining & Workflows

    Splitting work across steps, and knowing when one prompt is enough.

    7 prompts

  • Model Guides

    Where the models genuinely differ, and where the differences are overstated.

    7 prompts

  • Structured Output

    JSON, tables and schemas that validate on the first attempt.

    7 prompts

  • Troubleshooting

    Diagnosing the specific ways prompts fail, and what to change.

    7 prompts

  • Safety & Security

    Prompt injection, data handling and the failure modes that matter in production.

    7 prompts

A good first prompt from each

FundamentalsTurn a vague request into a prompt that worksFix a prompt that is producing generic output.Get the prompt →TechniquesGet the model to reason before answeringImprove accuracy on tasks that need working out.Get the prompt →System PromptsWrite a system prompt for a repeated taskSet up persistent instructions for an assistant or tool.Get the prompt →Chaining & WorkflowsDesign a multi-step prompt chainBuild a sequence of prompts that produce something one could not.Get the prompt →Model GuidesChoose the right model for a taskPick which model to use and why.Get the prompt →Structured OutputGet reliable JSON from a modelProduce output another system can parse.Get the prompt →TroubleshootingDiagnose why a prompt is not workingWork out what is wrong before changing anything.Get the prompt →Safety & SecurityAssess prompt injection risk in what you have builtWork out whether untrusted input can hijack your system.Get the prompt →

How to use these prompts

  1. Pick a category above, or search the whole prompt library.
  2. Press “Fill in” on any prompt to replace the [BRACKETED] parts in a simple form – or copy it and edit it in your AI tool.
  3. Read the tip under each prompt. It usually names the thing that goes wrong first.
  4. Got your own prompt? Run it through the Prompt Improver to see what is missing.