What this tool does
Prompt Variable Extractor scans reusable prompts for placeholders and returns a deduplicated variable inventory.
← All Tools
🔎 Prompt Variable Extractor
Paste a prompt template and get a deterministic inventory of placeholders like {{variable}}, {variable}, ${variable}, and [[variable]].
Supported placeholder styles
- Mustache variables such as
{{audience}}. - Single brace variables such as
{topic}. - JavaScript template variables such as
${tone}. - Double bracket variables such as
[[format]].
Why it is useful
It helps prompt engineers, agent builders, and template maintainers find required inputs before sharing or testing a prompt.
How it works
The browser applies deterministic placeholder patterns, keeps variables in first-seen order, counts repeats, and records line numbers.
Best input
Paste a reusable prompt template with clear placeholder names such as role, audience, topic, tone, or output_format.
Workflow tip
Copy the JSON summary into schema builders, template docs, or eval fixtures to make missing prompt inputs explicit.
Privacy note
Prompt scanning happens locally in your browser without model calls or server-side processing.
Important limitation
The extractor identifies placeholder syntax; it does not judge whether a variable is semantically required or safe to expose.
Quick answers
Does this call an AI model?
No. It uses deterministic browser-side pattern matching only.
Can it dedupe mixed syntaxes?
Yes. The same variable name found as {topic} and {{topic}} is counted as one variable with multiple occurrences.
Common Use Cases
Audit a shared prompt template
Extract the required variables before handing a prompt to teammates or an agent workflow.
Prepare prompt template docs
Copy the markdown table into documentation so every placeholder has an owner and expected value.
Build schema inputs from prompts
Use the JSON summary as a starting point for prompt schemas, structured-output prompts, or eval data fixtures.
AI Prompt Workbench workflow
Move between prompt optimization, reusable templates, variable extraction, few-shot examples, structured-output prompts, schema binding, agent planning, prompt diffs, and system prompt formatting without running an LLM.