Backend for a public web application with moderation of text messages

Technical task (TT)

Desired completion time - 24 hours (URGENT!)

Backend for a public web application with moderation of text messages

1. General description

It is necessary to develop a backend for a web application that operates in a single instance, located in a public space (screen / terminal / installation).

The client application sends short text messages (analogous to SMS), which undergo moderation and, after approval, are displayed on the screen.

The backend is installed on a separate PC and operates:

in a local network together with the client application;

or with access via the internet (optional).

2. Architectural and technological requirements

2.1 General requirements

The technologies must be widespread and easily supported.

The use of exotic or narrowly specialized frameworks that complicate the transfer of the project to another developer is prohibited.

The solution should be understandable to an external developer without prolonged onboarding.

2.2 Deployment and support

Mandatory:

the ability for quick deployment of the system;

the ability for simple updates.

Acceptable and desirable options:

storing code in GitHub;

deployment via Docker / Docker Compose;

minimal number of manual steps during updates.

The backend must run:

on a regular PC (Windows / Linux);

without dependence on specific server infrastructure.

3. Roles and access

3.1 Client application

Access:

sending messages;

receiving a list of approved messages;

receiving system status.

Authorization:

not mandatory or implemented through a simple mechanism (API-key, token).

3.2 Admin panel (moderation)

Single role — moderator.

Implementation of the admin panel — at the developer's choice:

separate web admin panel;

ready-made admin dashboard;

or another standard approach.

❗ The presence of a separate public API for the admin panel is not mandatory.

The main requirement is functionality and ease of support.

Moderator functions:

viewing all messages;

changing the status of a message;

editing text before approval;

managing the system state of the application.

4. Data and message model

4.1 Message

Field    Type    Description

id    string / UUID / int    Unique identifier

original_text    string    Original text from the client

approved_text    string | null    Text after moderation

status    enum    pending / approved / rejected

created_at    datetime    Date and time of receipt

reviewed_at    datetime | null    Date of moderation

ai_checked    boolean    Whether it has been checked by AI

ai_result    string | null    Comment or result of AI check

5. System state

5.1 SystemState

Field    Type    Description

is_blocked    boolean    Is the screen blocked

system_message    string    System message

updated_at    datetime    Date of last change

6. API for the client application (mandatory)

6.1 Sending a message

POST /api/messages

{

  "text": "Short text message"

}

{

  "success": true,

  "message_id": "12345"

}

6.2 Receiving approved messages

GET /api/messages/approved

{

  "messages": [

    {

      "id": "123",

      "text": "Approved or edited text",

      "created_at": "2026-01-08T10:15:00Z"

    }

  ]

}

Only messages with the status approved are returned.

6.3 Receiving system status

GET /api/system/status

{

  "is_blocked": true,

  "system_message": "Temporarily unavailable"

}

7. Automatic message checking via AI (optional)

Automatic checking of texts via OpenAI API.

The result of the check is stored and available to the moderator.

AI does not make the final decision, it only assists moderation.

8. Limitations and protection

Message length limitation (recommended up to 160–240 characters).

Basic protection against spam (rate-limit).

Validation of incoming data.

9. Expected result (MVP)

Working backend with a minimal API for the client.

Simple admin panel with one role.

Quick deployment on a PC.

The ability to easily transfer support to another developer.

Readiness to operate both in a local network and via the internet.