Accessibility Auditing
Checking HTML/React components for ARIA compliance and screen reader support.
Concrete prompting and workflow techniques for generating code with AI.
Checking HTML/React components for ARIA compliance and screen reader support.
Generating mock OpenAPI endpoints or JSON responses for frontend development.
Having the AI build the folder structure and empty files first.
Writing complex shell scripts by describing the workflow.
Asking the AI to think step-by-step before generating the code solution.
Generating GitHub Actions YAML specs for deployment automation.
Pasting git diffs and asking for conventional commit messages.
Asking the AI to break a monolithic React component into smaller units.
Feed the AI relevant code files or examples before generating.
Translating working code from one language to another.
Describing a layout visually to get the CSS Grid or Flexbox code.
Having the AI rewrite a Dockerfile to use multi-stage builds.
Writing the docstring and letting Copilot or an AI complete the function.
Pasting the entire stack trace for the AI to debug.
Providing 2-3 examples of desired input/output format before generating.
Generating a generic function, then asking for edge cases to be added.
Extracting hardcoded strings from components into i18n JSON files.
Having the AI summarize undocumented legacy spaghetti code.
Writing database schema migration scripts using ORM specific syntax.
Asking the AI to optimize an existing function for time or space complexity.
Writing out logic in English pseudo-code and having the AI compile it.
Providing examples of matching strings to generate complex regular expressions.
Asking the AI to act as a senior architect to review your design.
Explaining the problem to the AI to figure it out yourself or get a solution.
Using AI explicitly to audit code for OWASP vulnerabilities.
Writing a detailed specification document before generating code.
Describing data requirements in English to generate complex SQL joins.
Generating XState or standard state machine configs from descriptions.
Writing tests first, then making the AI write code that passes them.
Passing untyped Python/JS and asking for full strict typing.
Asking the AI to generate the boilerplate for mocking complex dependencies.
Asking for 10 possible naming conventions for a complex function.