AI Prompt Library: Practical Prompts by Task

Practical, searchable and reviewed by task

AI prompts you can adapt, test and improve

This is not a dump of one-line commands. Browse 1,052 prompt records by the work you need to do, replace the variables with real context, and check the output before you use it.

1
Task
2
Context
3
Constraints
4
Verify

Start with the outcome

Browse prompts by task

Role names are filters, not duplicate categories. Each prompt has one main home.

Use prompts as working documents

A prompt gets better when the evidence gets better

Before copying anything, decide what the model needs to know and what it must not invent. Replace placeholders, add examples, state the output format, then test the answer against the source material or a real tool.

  1. Define the task. Name the decision, document, analysis or artifact you need.
  2. Add real context. Give the model the audience, inputs, versions and constraints that matter.
  3. Set a quality check. Ask for assumptions, failure cases or a rubric that you can verify.
  4. Keep responsibility. Review code, facts and high-stakes decisions yourself.

Build the skill behind the prompt

Prompt guides from MetaCyberGuru

Editorial and safety policy

MetaCyberGuru excludes jailbreak and safeguard-bypass prompts. High-stakes medical, legal and financial records are not published without a safe, non-advisory rewrite and appropriate review. Prompt detail pages are noindex by default, which keeps the library useful without creating thousands of thin search pages.

Licensed source records retain internal provenance. Public guidance, category introductions and the professional code-review collection are written for MetaCyberGuru. Model names, tool capabilities and facts still need current verification.