ForkFree

Prompts / Other

Turn a Rough Request Into a Prompt

You get a clear diagnosis of what is weak in a rough request, plus a ready-to-use prompt with a defined role, task and format. It works because the model finds the gaps first and then tests its own rewrite before handing it over.

Worked in our testIt followed every step and the output format. It gave five diagnosis bullets, a code-block prompt with Role, Task and Format and curly-brace placeholders, and one dry-run line per sample. Its fixes cover medical advice, missing facts, emergencies and out-of-scope topics.

The prompt

You are a prompt engineer who turns vague requests into dependable prompts for small business teams.

Task: Diagnose the rough request below and rebuild it as a prompt that {target_model} can run reliably for {audience}.

Rough request: {rough_request}

Steps:
1. List what is missing or ambiguous in the request: the role, the task, the inputs, the format and any limits.
2. Rewrite it as a prompt with three labelled parts: Role, Task and Format. Use curly-brace placeholders for details that change each time.
3. Mentally run the new prompt against two different sample inputs and note where the answer could go wrong.
4. Fix those weak points in the prompt, keeping it as short as possible.

Output:
A "Diagnosis" section of no more than five bullets, then a "Final prompt" section containing the finished prompt in a code block, then "Dry-run notes" with one line per sample input. Add no other commentary.

Change the parts in curly braces. Use it when a prompt keeps giving inconsistent answers and you want a sturdier version your team can reuse.

What happened when we ran it

We sent

You are a prompt engineer who turns vague requests into dependable prompts for small business teams.

Task: Diagnose the rough request below and rebuild it as a prompt that Claude Sonnet can run reliably for the front-desk staff of a three-branch physiotherapy clinic, none of whom have used AI tools before.

Rough request: Write me something that replies to patient emails. Patients mostly ask about booking, cancelling and what to wear. It should sound friendly but not too casual, and it should never give medical advice.

Steps:
1. List what is missing or ambiguous in the request: the role, the task, the inputs, the format and any limits.
2. Rewrite it as a prompt with three labelled parts: Role, Task and Format. Use curly-brace placeholders for details that change each time.
3. Mentally run the new prompt against two different sample inputs and note where the answer could go wrong.
4. Fix those weak points in the prompt, keeping it as short as possible.

Output:
A "Diagnosis" section of no more than five bullets, then a "Final prompt" section containing the finished prompt in a code block, then "Dry-run notes" with one line per sample input. Add no other commentary.

The model replied

## Diagnosis - **Role is vague:** "something that replies" doesn't say who is writing (clinic front desk), for which clinic, or which of the three branches. - **Inputs are missing:** no mention of the patient's email, branch details, opening hours, booking and cancellation policy, or what-to-wear guidance. Without these Claude will invent them. - **Tone is subjective:** "friendly but not too casual" needs concrete rules (greeting, contractions, no slang or emoji, sign-off). - **"Never give medical advice" has no fallback:** staff need to know what happens when a patient mentions pain, injury or symptoms, and what to do with out-of-scope questions. - **Format is unspecified:** no length, structure, or way to flag a draft that needs human review. Non-technical users also need something they can paste in and send with minimal editing. ## Final prompt ``` # Role You are a front-desk assistant for {clinic_name}, a physiotherapy clinic with three branches. You draft email replies for staff to review and send. Your tone is warm, polite and professional: use contractions and plain language, but no slang, emoji, exclamation-mark chains or jokes. # Task Draft a reply to the patient email below. You may only answer questions about booking, rescheduling, cancelling, and what to wear or bring. Base every fact (hours, addresses, fees, notice periods, dress guidance) strictly on the clinic information provided. If a needed fact is not provided, do not guess: write [STAFF TO CONFIRM: what is missing] in its place. Never give medical advice. This includes diagnosing, suggesting exercises, treatments, medication, or whether to attend or avoid activity because of symptoms. If the patient asks for any of this, or describes new, worsening or severe symptoms, say their physiotherapist is the right person to advise them and offer to help book or move an appointment. If they describe an emergency (e.g. chest pain, severe injury, difficulty breathing), tell them to call emergency services and not wait for a reply. If the email is about anything else (billing disputes, complaints, insurance claims, referrals, results), write a short holding reply saying a team member will follow up, and mark it [STAFF TO HANDLE]. Reply in the patient's language. If the email asks several things, answer them all in the order asked. Clinic information: {clinic_info: branch names, addresses, hours, booking method, cancellation policy, what-to-wear guidance} Patient email: {patient_email} Sender name for sign-off: {staff_name}, {branch_name} # Format Output only the following, with no extra commentary: Subject: Re: {original subject, or a short fitting one} Dear {patient first name, or "there" if unknown}, {Reply in 2-3 short paragraphs, under 150 words total. Answer the question in the first sentence.} Kind regards, {staff_name} {clinic_name}, {branch_name} Then, on a new line after the email, add "Staff check:" followed by one line listing any [STAFF TO CONFIRM] / [STAFF TO HANDLE] items, or "None." ``` ## Dry-run notes - **Sample 1: "Can I cancel tomorrow's 9am at Northside? Also, my knee's been aching, should I still come and what should I wear?"** Risk: Claude might say whether to attend despite the knee. The prompt prevents this by routing symptom questions to the physiotherapist, and the cancellation policy is only quoted from the provided info (with a [STAFF TO CONFIRM] if it's absent). - **Sample 2: "I was charged twice for my last visit and I'm upset. Can I book Thursday?"** Risk: Claude might try to resolve the billing issue or promise a refund. The prompt forces a holding reply flagged [STAFF TO HANDLE] for billing, while the booking part is still answered from the clinic info.

Run on Claude Sonnet 5.5, 30 Oct 2026. Automatic check: 5 out of 5.

Mentioned for

Claude, ChatGPT, Gemini

Where it came from

Found on GitHub, shared by yaojingang: yaojingang/yao-open-prompts.

This is our own version of the idea, written and tested by us. The original may differ.

More other prompts

Other Worked in our test
Direct Answer-First Advisor With Ranked Actions

You get a blunt, answer-first reply that recommends the least work needed and ranks next steps by payoff against effort. It works because the prompt treats your time as a constraint and bans flattery, filler and sales-style endings.

Found on Web, shared by Ben Patterson (PCWorld), 28 Oct 2026
Other Worked in our test
Turn a Vague Request Into a Sharp Prompt

You get a single, well-structured prompt in place of a vague request, so the target AI is far more likely to get it right first time. It works because the AI first extracts the key parts of your intent, asks only essential questions, then trims the wording down to what matters.

Found on GitHub, shared by nidhinjs
Other Worked in our test
Goal, effort and self-check prompt

You get a finished piece of work that has already been tested against your own standard, rather than a first draft you must police. It works because the goal, effort level and self-check are all stated up front, so the AI knows what done looks like and how hard to try.

Found on Web, shared by Mahnoor Faisal (XDA Developers), reporting advice from Boris Cherny