中文

Structured Prompt Builder

Build deterministic prompts from typed fields, CO-STAR or RTF structures, ordered examples, and explicit local templates.

Deterministic local generation

Structured prompt workspace

No model calls · No default body persistence
Canonical field order with no prefilled content.
The template controls field order and labels; this independently controls output syntax. Copy buttons can also choose a format directly.
Examples (optional)

Add input/output examples in order. Blank examples are omitted; move buttons provide keyboard-accessible reordering.

Live plain-text preview

Generated prompt

Plain
Fill any field to generate a prompt here. Empty fields never create headings or tags.
0 UTF-16 characters— Token
Exact for models using this encoding; an estimate otherwise. Chat overhead excluded.
Explicit local saves

My templates

0 / 40

Only “Save template” or importing templates writes field bodies to this browser’s localStorage. Unsaved drafts remain in page memory only; clearing saved templates never resets the draft.

or drop a local file here (up to 1 MB)
Generation, copying, tokenization, import, and export run locally; prompts are never sent to an online model.
About this tool

Made for the task at hand.

Turn prompt requirements into a repeatable structure without calling a model. Typed fields, stable example ordering, and independent output serialization make every result deterministic, while explicit local saves keep reusable templates under your control.

How to use

  1. Choose Blank, CO-STAR, RTF, Markdown, or XML structure, then fill any fields and add or reorder examples. Structure templates control ordering and labels; the output selector independently controls syntax.
  2. Review the live preview and real token count for the selected encoding. Copy plain text, Markdown, or escaped XML, or download the current preview.
  3. Explicitly save, load, rename, duplicate, import, export, delete, or clear browser-local templates. Reset asks for confirmation and never deletes saved templates.
Your work stays yours.
Tool inputs are processed locally in your browser and are never sent to KitVerse servers. Clear the workspace when using a shared device.
Common questions
Does this tool generate prompt content with AI?

No. It only assembles the text you enter using deterministic field ordering. There are no backend requests or online model calls.

How are structure templates different from output formats?

CO-STAR and RTF change field order and structural labels. Plain text, Markdown, and XML are independent serializers, so you can copy any structure in any supported syntax. Empty fields and blank examples are omitted.

Are drafts and templates saved automatically?

Drafts remain in page memory only and disappear on refresh. Only Save template or a confirmed import writes field bodies to this tool’s separate localStorage namespace. Delete and Clear saved templates do not clear the current draft.

What do token counts and limits mean?

Real o200k_base or cl100k_base counts are exact for models using that encoding and estimates otherwise, excluding chat overhead. Each field allows 100,000 UTF-16 characters, with up to 40 examples and 40 saved templates. Imports are files up to 1 MB and versioned stores up to 4,000,000 characters, subject to browser quota.

Related tools · AI & LLM

New

Tokenizer Playground

See exactly how text becomes tokens, then compare the cost across popular LLMs.

Free

Token Counter

Count characters and LLM tokens with a lightweight, private text workspace.

Free

Token Cost Calculator

Estimate input and output spend with transparent per-million-token pricing.