中文

Prompt Batch Variable Expander

Expand {{variable}} prompts from JSON or CSV records with warnings, paginated previews, and lossless exports.

Deterministic batch variable expansion

Prompt batch variable workspace

No model calls · No body persistence
4 variables
{{tone}}{{audience}}{{project.name}}{{note}}
or drop a local file here (up to 1 MB)
≤ 200
186 / 1,000,000 charsDuplicate CSV headers are rejected, never overwritten
or drop a local file here (up to 1 MB)

Unknown placeholders are preserved and warned about by default; present empty values are distinct from missing values.

Bounded synchronous processing · 10 records per page
Enter a template and variable data, then generate a single or batch preview.
Up to 200 records; larger input asks you to split batches and is never silently truncated. Every export includes all results.
About this tool

Made for the task at hand.

Turn one reusable prompt template into a reviewable batch without calling a model. Direct keys and nested dot paths follow the existing Prompt Template Formatter semantics, while detailed diagnostics keep missing data distinct from values that are present but empty.

How to use

  1. Enter a template with {{variable}} placeholders and provide one JSON object, an array of objects, or CSV records up to 1 MB.
  2. Generate single or batch previews, review missing and extra fields plus empty and null values, and choose whether unknown placeholders remain visible or become empty strings.
  3. Select and copy individual records, page through results 10 at a time, or export every accepted result as JSONL, correctly quoted CSV, or TXT.
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
How are duplicate CSV columns handled?

Duplicate headers are rejected before records are constructed, and the duplicate names are reported. The tool never silently keeps only the first or last value. JSON objects already have unique keys after parsing, so no synthetic duplicate detection is applied there.

What are the batch and display limits?

Up to 200 records and 1,000,000 UTF-16 data characters are accepted. Larger batches must be split; nothing is silently truncated. Generation is bounded synchronous work and only 10 record summaries plus one selected full preview are rendered at once.

Are empty and missing values the same?

No. An empty string and null are present values and are reported separately. Missing or undefined paths are unknown placeholders: they remain unchanged with a warning by default, or become empty strings only when you select that option.

Is prompt data saved or sent anywhere?

No. Templates and records stay in page memory, with no body persistence, backend requests, or model calls. Imports and exports use local browser file APIs only.

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.