HomeCasa Prompts

Improve an existing prompt: diagnosis, rewrite and test cases

Paste a prompt that gives weak results and get a diagnosis of its flaws, a rewritten and structured version, a change log and test cases to check it really works better. Designed for ChatGPT, Claude, Gemini and Microsoft Copilot. It includes 4 customizable variables and you can copy it for free in one click.

Advanced · Text · Pending editorial review

Designed for:

Prompt

Act as a prompt engineer who audits and rewrites instructions for large language models.

My current prompt:
"""
{{original_prompt}}
"""

What I want it to achieve: {{goal}}
Model I use: {{model}}
What goes wrong today: {{problem}}

Deliverables:
1. Diagnosis: list the concrete weaknesses (ambiguous instructions, missing context, no output format, conflicting rules, no examples, no guidance for missing information). Quote the exact fragment for each one.
2. Questions: up to 5 questions whose answers would improve the prompt the most. If one is essential, ask me before rewriting; otherwise state the assumption you are making.
3. Rewritten prompt in a code block, structured as role, context, task, numbered steps, constraints and output format. Keep any placeholders or variables I already use.
4. Change log: a table with change, reason and expected effect.
5. Test set: 5 varied inputs (including one edge case and one off-topic or adversarial input) and, for each, what a good answer must contain.
6. A short version (under 80 words) for quick use.

Constraints:
- Do not add facts about my business or product that are not in the original; mark gaps as [FILL IN].
- Do not promise the new version will perform better: explain why it should and how to verify it with the test set.
- Adapt to {{model}} only where it really matters (for example, tags to delimit data or splitting system and user instructions).

Format: one heading per deliverable and the rewritten prompt in a single copy-ready block.

How to use it

Paste the prompt exactly as you use it today and describe, with a real example, the answer that disappointed you. Then run the 5 test inputs through both the old and the new version and compare before you switch.

Usage example

Prompt: 'Write a LinkedIn post about our software' · Goal: posts with a hook and no ad tone · Model: ChatGPT · Problem: they come out generic and full of emojis.

Sample output

1. Diagnosis

  • "Summarize this document": no length limit, so the model decides and runs long.
  • "tell me what matters": ambiguous; it never says decisions, risks and figures are what matter.
  • No output format and no rule for reports that lack data.

2. Assumption

I'm assuming reports are pasted as plain text and the reader is leadership. Correct me if not.

3. Rewritten prompt

Role: analyst who summarizes internal reports for leadership.
Task: summarize the report inside <report></report> in at most 1 page (~350 words).
Steps:
1. List decisions made or pending.
2. List risks with their impact.
3. Copy the key figures verbatim (amounts, deadlines, %).
Constraints: add no outside facts; if something is missing, write [FILL IN].
Format: Decisions · Risks · Key figures · Next steps.

4. Change log

ChangeReasonExpected effect
350-word limitSummaries too longShorter output
"Key figures" sectionNumbers were droppedFigures always shown

5. Test set (excerpt)

  1. Report with no figures → must say "no figures given" instead of inventing them…

Key facts

Category
Programming & development · AI agents
Output type
Text
Level
Advanced
Designed for
ChatGPT, Claude, Gemini and Microsoft Copilot
Verified in
Pending editorial verification
Variables
Current prompt, What you want to achieve, Model and What goes wrong today
Language
English
Price
Free, no sign-up
Updated
Nov 7, 2026

Frequently asked questions

Why does it ask for test cases instead of just rewriting?

A prompt is only better if it gives better answers on real inputs. Running both versions on the same inputs stops you from keeping the one that merely looks more professional.

Does it work for chatbot or agent system prompts?

Yes. Say in the goal that it is a system prompt and add typical user questions plus attempts to bypass the rules to the test inputs.

What is this prompt for?

Paste a prompt that gives weak results and get a diagnosis of its flaws, a rewritten and structured version, a change log and test cases to check it really works better.

Which AI tools does it work with?

It is designed for ChatGPT, Claude, Gemini and Microsoft Copilot. We have not verified it in these tools yet, so results may vary between them.

How do I customize the prompt?

Click “Customize” and fill in the variables (Current prompt, What you want to achieve, Model and What goes wrong today). The text updates instantly and you can then copy it in one click. Values are stored only in your browser.

Do I need experience to use it?

It is advanced: it is meant for people with experience in the topic who can evaluate and adjust the result.

Is this prompt free?

Yes. Every HomeCasa Prompts prompt is free, with no sign-up or subscription. You can also use them in professional projects.

Can I trust the AI output?

Treat it as a high-quality draft: always check data, figures and important claims before publishing or making decisions.

Related prompts