Prompt Generator Guide
Start with an observable deliverable
The generator structures a request; a separate assistant supplies the answer. This walkthrough turns a vague workshop request into a checklist instruction. Success means an answer with the required items and no invented event details, not simply a longer prompt.
Step 1: replace the vague task
Before: “Help with the workshop.” After: “Write a packing checklist for a drawing workshop.” The revised task identifies an artifact and an activity. Put it in Task; put “Attendees bring their own pencils.” in Context so the assistant need not guess who supplies them.
Step 2: inspect the generated request
For the reproducible run, use the English field values below and select Concise. Generate and compare the output character for character. The task and context remain intact. The generic closing still requests a thorough answer, so the style does not enforce a length limit.
Input and selections
Task: Write a packing checklist for a drawing workshop.
Context: Attendees bring their own pencils.
Style: Concise
Exact tool output
You are an expert assistant. Be brief and to the point. Focus only on essential information.
Task: Write a packing checklist for a drawing workshop.
Context: Attendees bring their own pencils.
Please provide a thorough and helpful response.Step 3: define a useful revision
If the prompt needs a tighter boundary, revise the task to “Write a packing checklist for a drawing workshop. Use five bullets and distinguish attendee items from organizer items.” Generate again. This is a deliberate change to the request, not a claim about what an assistant will return.
Step 4: evaluate the answer elsewhere
Copy the request to your chosen assistant. Check that pencils are assigned to attendees and that the list respects the requested structure. If either check fails, revise that one requirement. Keep the working prompt and an example answer together for later comparison.
Mistake 1: looking for a model selector
There is no model selector on this page. The style buttons insert fixed sentences; they do not tune an AI model. Choose the destination service separately.
Mistake 2: adding unsupported background
More context is useful only if it is relevant and true. Do not add invented venue facilities to make the request sound complete; an assistant might turn those details into requirements.
Worked example: omit optional context
This orientation request uses Professional and leaves Context empty. The generator omits the entire Context line instead of filling it with a guess. Notice that the requested three questions remain part of Task; the style adds only its fixed business tone sentence. The output is still a request, so no actual orientation questions appear here.
Input and selections
task: List three questions for a volunteer orientation.
context: (empty)
style: professional
Exact tool output
You are an expert assistant. Use formal language and maintain a business-appropriate tone.
Task: List three questions for a volunteer orientation.
Please provide a thorough and helpful response.Worked example: bound a creative task
For a seed exchange, Creative changes the opening instruction while Context carries a restriction on claims. Compare the two generated requests: neither the role sentence nor the closing is tailored to the topic. If an outside assistant suggests an exaggerated name, strengthen the task or context yourself; generating again with unchanged fields will not add new safeguards.
Input and selections
task: Suggest two names for a neighborhood seed exchange.
context: Avoid promises about crop yields.
style: creative
Exact tool output
You are an expert assistant. Think outside the box. Provide innovative and unique perspectives.
Task: Suggest two names for a neighborhood seed exchange.
Context: Avoid promises about crop yields.
Please provide a thorough and helpful response.Mistake 3: expecting a separate format setting
There is no dedicated output format control. Put a table, bullet count, or required fields into Task or Context; the generator simply preserves those words in its template.
Mistake 4: leaving the task blank
Context alone does not trigger generation. A blank or whitespace-only task returns immediately, so make sure the deliverable is in Task before interpreting a displayed result as current.
Mistake 5: treating regeneration as variation
The same inputs and style produce the same string. To compare alternatives, change a field deliberately and keep a record of which requirement changed.
Tips for better results
- Put the deliverable first because Task appears before Context.
- Use Context for facts that qualify the task, and leave it empty when unnecessary.
- Read the fixed style sentence and closing together before copying the assembled request.