Claude Prompt Generator

Describe your goal in plain English. PromptVibe writes a Claude prompt structured the way Anthropic recommends — each part in its own XML tag, long material first, instructions spelled out. Free, no account needed.

A Claude prompt generator turns a plain-language goal into a prompt structured the way Anthropic's documentation recommends: clear, explicit instructions with the reason behind them, long documents placed above the question, each part wrapped in its own XML tag, and a few examples of the output you want. PromptVibe writes that structure for you, free.

Contract review — long document first, quotes before analysis

<contract>
[Paste the full contract text here]
</contract>

<instructions>
You are reviewing this contract for a small agency that is the service provider.
1. Copy the clauses about payment terms, liability and termination into <quotes> tags, word for word.
2. Using only those quotes, rate each clause Low / Medium / High risk for the agency, with one sentence on why.
3. End with the three points the agency should negotiate, in plain English.
If one of these clauses is missing, say so instead of guessing.
</instructions>

Customer support reply — role + tagged example

<role>
You are the support lead at a small software company. Customers read replies on their phones, so keep them short and warm.
</role>

<customer_email>
[Paste the customer's email here]
</customer_email>

<task>
Reply to the email above: acknowledge the problem in one sentence, give the fix as numbered steps, and offer a direct reply if it does not work. Under 120 words.
</task>

<examples>
<example>
Sorry the export failed twice — that is frustrating. Here is the fix:
1. Open Settings → Data.
2. Choose "Export as CSV" instead of XLSX.
If that does not work, reply here and I will run the export for you.
</example>
</examples>

Code change — explicit action + output format

<context>
Next.js 14 app in TypeScript. On slow connections the checkout button can submit twice.
</context>

<task>
Change the CheckoutButton component so it cannot submit twice: disable it while the request is pending and re-enable it if the request fails. Make the change — do not just describe it. Keep the existing props and styling.
</task>

<output_format>
The full updated component in one code block, then two sentences on what changed.
</output_format>

• PromptVibe is independent and not affiliated with or endorsed by Anthropic.
• The example prompts on this page are illustrative. They show structure; they were not benchmarked.
• A well-structured prompt improves the odds, not the outcome. Test it on your real task before relying on it.
• Claude behaves differently across model versions and between the Claude apps and the API. Prefill and similar API details do not apply in the apps.

Is this an official Anthropic or Claude tool?

No. PromptVibe is an independent tool with no affiliation with Anthropic. It generates structured prompts you can use with Claude, but it is not created or endorsed by Anthropic.

Should I use XML tags in Claude prompts?

For anything longer than a quick question, yes. Anthropic's documentation recommends XML tags when a prompt mixes instructions, context, examples and inputs. Use clear, consistent names such as <instructions>, <context> and <input>, and refer to them by name in your instructions.

Where should I put a long document in a Claude prompt?

At the top, above your question and instructions. Anthropic recommends this placement for long inputs. For long documents, also ask Claude to quote the relevant passages first and answer from those quotes.

Do I still need to tell Claude to "think step by step"?

The phrase on its own adds little with current models, which can reason before answering. A stronger instruction is to state the criteria a good answer must meet and ask Claude to check its answer against them before it finishes.

Do the PromptVibe frameworks work with Claude?

Yes. All seven — Standard, ROLE, TAG, COAST, RACE, RISE and APE — work with Claude. When Claude is the target model, PromptVibe writes the result with XML tags so the role, task and constraints stay separate.