Give each thing
one name.
If the component is a “worker,” keep calling it a worker. A fresh synonym can look like a different component.
Experimental
Better prompts.
Clearer answers.
Describe your task. Choose the shape of the answer. Copy a complete prompt into your preferred AI tool.
Describe your task, the result you need, and any important constraints.
Comparing websites or products?
Included exactly as entered. If your AI cannot open links, paste the relevant text here.
Built in your browser and never sent anywhere.
Start with your task.
Complete these details to copy:
Your prompt will appear here as you type. Add your task and choose the answer you want.
01 DESCRIBE02 CHOOSE03 COPY & PASTE
CONSISTENT TERMS✳EVIDENCE VS. INFERENCE✳CAVEATS KEPT✳ASK, DON’T GUESS
CHOOSE THE SHAPE OF THE ANSWER
A little structure.
A clearer result.
An explanation, steps, a table, a diagram, or a small tool: pick the form in the creator above. Press Chaos to see the same points without structure.
WHAT THE PROMPT ASKS FOR
Good technical writing
leaves less to guess.
Fluent is not the same as clear. Ask for stable names, direct sentences, and an explicit boundary between evidence and inference.
Give each thing
Give each step
one action.
Break a procedure into actions you can perform and check. Put conditions before the actions they control.
Give each claim
a clear status.
Separate what the source says from what the model infers. Keep uncertainty visible. Ask how to verify the answer.
Same intent. Less interpretation.
After saving the configuration, you might want to restart the service and check that the worker’s health endpoint returns HTTP 200, restoring the previous configuration if the process fails this check.
A crowded sentence. Inconsistent names.
- Save the configuration.
- Restart the worker.
- Check the worker’s health endpoint.
Expected result: HTTP 200. If the check fails, restore the previous configuration.
One name. Visible steps. A defined result.