dotpmt

a super simple way to separate your prompts from your code.

When working with LLMs (like OpenAI, Claude, or open-source models), prompts often grow into long, structured text blocks — but they're usually hardcoded as strings inside your codebase.

This quickly becomes messy, hard to maintain, and makes versioning prompts painful.

dotpmt solves this by letting you:

• Keep prompts in their own .pmt files
• Use variables like {{input}} for lightweight templating
• Treat prompts like real content — editable, diffable, and readable

A .pmt file is just a plain text file with an optional templating syntax:

# summarize.pmt
Summarize the following:

{{input}}

You can store .pmt files anywhere in your project, commonly in a prompts/ folder.

import { loadPrompt } from "dotpmt";

const prompt = loadPrompt("prompts/summarize.pmt", {
  input: "YOUR LONG TEXT",
});

console.log(prompt);