SOFT CAT.ai
FIND SOMETHING USEFUL
← Prompt recipes

PROMPT ENGINEERING / A WORKED STARTING POINT

Shorten a prompt without losing a rule

Compress repeated wording while preserving explicit constraints and showing how to check the revision.

Template reviewed

WHEN TO USE IT

A working prompt repeats itself and you want a reviewable edit.

Bring these inputs

{{original}}
The complete prompt, including constraints and any abstention rule.

01 / THE PROMPT

A template you can inspect.

Shorten the prompt below without changing its task, required output or boundaries.

Return the revised prompt, then a compact requirement map showing where every original rule survives. Do not add new constraints or remove an abstention rule to save words. If rules conflict, identify the conflict instead of silently choosing one. Suggest two small test inputs that could expose a lost requirement. Do not promise that a shorter prompt will improve answer quality.

Original prompt:
{{original}}

Copying does not run the prompt. Workbench opens an editable draft with the example inputs. It does not save or send it.

02 / THE EXAMPLE TARGET

What a useful result should contain.

This is a target for the worked example, not a recorded model response. Equivalent wording can be valid where the task allows it.

Summarise the supplied source in exactly three bullets, with no introduction or conclusion. Use no outside facts. Write UNKNOWN for any requested point the source cannot answer.

03 / JUDGE THE RESULT

Check the answer, not the confidence.

  • Exactly three bullets, no surrounding prose and source-only answers remain required.
  • The UNKNOWN rule survives unchanged in meaning.
  • The requirement map identifies both duplicate and unique rules.
Open the related workshop tool ↗

The worked example is a target to inspect, not a saved response from a model. Use the checks to judge an actual result.

This template was revised during the September prompt review. The earlier text remains in Git history.

Read the prompt collection review ↗