How do template variables work?

Add key=value lines in the template variables field; matching placeholders in the prompt are replaced, and a table shows which variables were applied.

Can I target a specific model like ChatGPT or Claude?

Yes. The target model selection labels the prompt for ChatGPT, Claude, Gemini, or a general LLM, while the role, task, context, and constraints structure stays the same.

Is my prompt content sent to an AI service?

No. The prompt is assembled entirely in your browser and nothing is uploaded — you copy it into the model you use.

What makes a prompt work better?

Be specific about the task, give the context the model cannot know, state constraints such as length or what to avoid, and describe the output format exactly, for example "a Markdown table with columns Risk, Evidence, Fix". Separating facts from assumptions also makes answers easier to check.

AI prompt generator

A structured prompt gives a language model a role, a task, the context it needs, constraints, and the exact output format, so answers are more consistent.

Build structured prompts for ChatGPT, Claude or Gemini from templates, with role, context, constraints, variables and output format. Runs in your browser.

Browser-sideNo accountCopy, download, or print

Last reviewed 30 Oct 2026

Your inputs are saved as a draft in this browser only. Nothing is uploaded.

Jump to result

Result

TemplatecodeReview

Target modela general large language model

Estimated tokens138

Word count75

Review and adapt the generated prompt before sending it to a model.

Variables applied

Variable	Value
language	TypeScript
focus_areas	security, performance, maintainability
audience	engineering leaders
industry	enterprise software

[Target model: a general large language model]

You are Senior enterprise API architect.

Perform an expert code review. Report a quality assessment, likely bugs and security issues, performance optimisations, best-practice fixes, and refactoring opportunities.

Task:
Review the API design and identify governance

Context:
Source systems include SAP ERP

Constraints:
Avoid unsupported claims. Separate facts

Output format:
Markdown sections with a risk table and next actions

Write in a professional tone. Keep the response moderately detailed.

[Target model: a general large language model]

You are Senior enterprise API architect.

Perform an expert code review. Report a quality assessment, likely bugs and security issues, performance optimisations, best-practice fixes, and refactoring opportunities.

Task:
Review the API design and identify governance

Context:
Source systems include SAP ERP

Constraints:
Avoid unsupported claims. Separate facts

Output format:
Markdown sections with a risk table and next actions

Write in a professional tone. Keep the response moderately detailed.

How to use it

Pick a template and target model, then set the tone and response length.
Fill in the role, main task, context, and constraints.
Add template variables as key=value lines to replace placeholders.
Generate the prompt, check the estimated token count, and copy it into your model.

Questions

How do template variables work?: Add key=value lines in the template variables field; matching placeholders in the prompt are replaced, and a table shows which variables were applied.
Can I target a specific model like ChatGPT or Claude?: Yes. The target model selection labels the prompt for ChatGPT, Claude, Gemini, or a general LLM, while the role, task, context, and constraints structure stays the same.
Is my prompt content sent to an AI service?: No. The prompt is assembled entirely in your browser and nothing is uploaded — you copy it into the model you use.
What makes a prompt work better?: Be specific about the task, give the context the model cannot know, state constraints such as length or what to avoid, and describe the output format exactly, for example "a Markdown table with columns Risk, Evidence, Fix". Separating facts from assumptions also makes answers easier to check.