Zero dependencies

Prompt Runtime

Update prompts without redeploying. The promptodex npm module fetches and renders prompts at runtime—decoupling your AI logic from your release cycle.

$ npm install promptodex

The Problem

Hardcoded prompts create deployment bottlenecks

Without promptodex

1Prompts hardcoded in source code
2Small tweak requires full deploy
3Wait for CI/CD pipeline
4Prompts scattered across services
5Versioning via Git is clunky

With promptodex

1Prompts hosted on Promptodex
2Update prompts instantly
3No code changes needed
4Single source of truth
5Built-in versioning & rollback

See it in action

app/api/summarize/route.ts

import { pod } from "promptodex";
import { openai } from "@/lib/openai";

export async function POST(request: Request) {
  const { content } = await request.json();
  
  // Fetch prompt at runtime - update anytime without redeploying
  const prompt = await pod("summarize-article", {
    content,
    maxLength: "200 words"
  });

  const response = await openai.chat.completions.create({
    model: "gpt-4.1",
    messages: [{ role: "user", content: prompt }]
  });

  return Response.json({ summary: response.choices[0].message.content });
}

The summarize-article prompt lives on Promptodex.
Update it anytime—your API uses the new version immediately.

Built for real workflows

Three patterns that make teams more productive

Prompt Hosting

Host prompts on Promptodex and fetch them at runtime. Your prompt engineers can iterate independently without waiting on engineering deploys.

const prompt = await pod("customer-support-reply", {
  customerName: ticket.name,
  issue: ticket.description,
  tone: "professional"
});

Shared Fragments

Use the same prompt across multiple services. Web app, mobile backend, internal tools—all pulling from one source of truth.

// service-a
const prompt = await pod("company/extract-entities", { text });

// service-b  
const prompt = await pod("company/extract-entities", { text });

// Update once, both services get the change

Version Pinning

Pin production to tested versions while iterating on new ones. Roll back instantly if something goes wrong.

// Production: pinned to version 5
const prompt = await pod("code-review@5", { code, language });

// Staging: always latest
const prompt = await pod("code-review", { code, language });

Simple API

Three functions, zero configuration

pod(slug, variables?, options?)

Fetch and render a prompt in one call. The most common way to use promptodex.

Returns: Promise<string>

const prompt = await pod("greeting", { name: "World" });
// "Hello World, welcome to Promptodex!"

fetchPrompt(slug, options?)

Fetch the raw template without rendering. Useful when you need the template content itself.

Returns: Promise<PromptResponse>

const { content } = await fetchPrompt("greeting");
// "Hello {{name}}, welcome to Promptodex!"

renderPrompt(template, variables?)

Render a template locally. No network call—useful when you already have the template.

Returns: string

const rendered = renderPrompt("Hello {{name}}!", { name: "World" });
// "Hello World!"

Features

Everything you need to manage prompts at runtime

Runtime fetching

Prompts are fetched from Promptodex at runtime, not bundled at build time

Template rendering

Use {{variable}} syntax to inject dynamic values into your prompts

Version support

Pin to specific versions with @1, @2, etc. or always get the latest

Private prompts

Access private prompts securely with API key authentication

Zero dependencies

Under 200 lines, uses native fetch. No bloat, no security concerns

TypeScript ready

Full type definitions included. Great editor support out of the box

Private prompts for teams

Keep proprietary prompts secure while still getting the benefits of runtime hosting

Using private prompts

import { pod } from "promptodex";

// Store your API key securely
const PROMPTODEX_KEY = process.env.PROMPTODEX_API_KEY;

// Access private prompts with your API key
const prompt = await pod(
  "internal/sales-email-template",
  { 
    customerName: lead.name,
    product: lead.interest,
    repName: rep.name
  },
  { apiKey: PROMPTODEX_KEY }
);

Ready to decouple your prompts?

Install the promptodex module and start hosting prompts at runtime.

$ npm install promptodex