Coding · Excerpt

family-messenger master prompt and five steps

A long Russian spec plus five step prompts that guide an AI coding agent to build a four-platform family messaging app from scratch.

Author
Evgeny Khramov(GitHub 用户 hram) ↗
Models / tools
Not specified by the author
License
MIT ↗
Original language
ru
Нужно с нуля создать MVP self-hosted семейного мессенджера для связи родителя с детьми.
Цель: надёжная переписка в условиях нестабильной доступности публичных мессенджеров.
Результат должен быть в виде реального проекта с каталогами, кодом, SQL-схемой, документацией, docker-конфигурацией и инструкцией запуска.
Read the full prompt in the author’s original post ↗

Only an excerpt is shown here; full-text republication permission has not been verified.

Attribution required — keep the author’s name and original source when reusing this prompt. Author: Evgeny Khramov(GitHub 用户 hram) | Original source ↗

How to use it

VariableWhat to enterExample
项目目的和用户角色Project purpose and user roles—
目标平台Target platforms—
部署限制Deployment constraints—
需要的功能Required features—
各阶段的数据契约Data contracts for each step—

Models / tools: Not specified by the author

  1. Read the master prompt first and replace the purpose, platforms, deployment limits and features with your own, then give the five step prompts to a coding agent in order.

What it can do

This is a set of prompts shared by Evgeny Khramov (GitHub user hram) for having an AI coding agent build a self-hosted family messaging app from scratch, so parents and children can stay in touch when public messengers are unreliable. The master prompt is a long Russian spec that sets out the architecture, the data contracts shared by backend and clients, the API, the folder structure and the acceptance requirements. Five step prompts then cover the shared module, backend, deployment, cross-platform clients and a final self-review, in that order. The author says on DEV that he wrote none of the app code himself and had a first working version in about three days, running on Android, iOS, desktop and web. What sets it apart is that the prompt is a full engineering spec, showing how a long prompt can keep a multi-platform project on track. Read the master prompt first and replace the purpose, platforms, deployment limits and features with your own, then give the five step prompts to a coding agent in order.

Who it is for

Developers using AI coding agents on bigger projects who want to learn how to turn requirements into a spec. If you don't write code, you won't need it.

Before you use it

The original is in Russian with English technical terms. The author doesn't say which coding agent ran it. The resulting project runs on your own server and needs some Docker and Kotlin know-how. The security and privacy requirements in the prompt are design goals, not proof that the finished app is safe, so check it yourself before children use it. Your code and project material go to whichever model service you choose. The prompts are released under the MIT licence, so you can share and adapt them, but a full copy must include the author's copyright notice and the licence text; for now we show only an excerpt.

Follow the stated license. Do not use the output to replicate someone else’s work or a real person’s likeness.

Author examples

Based on the author's public posts. Not tested by AI Inspiration World.