malten

README

Malten

Anonymous ephemeral messaging

Overview

Malten is an ephemeral messaging app. It contains solely streams of text and nothing more. Streams have an idle lifetime of 1024 seconds. Each stream supports 1024 messages as a FIFO and 1024 characters per message. Streams can be discovered through exploration or via the API.

Design

Malten keeps everything in memory, nothing is ever written to disk or a database. This is to ensure privacy and security. We do not want to persist data and ideally also want to encrypt messages on the client side. Streams are aged out based on TTL and each stream limits the number of messages and char length to ensure we can comfortably run malten in memory on most devices.

Roadmap

• Presence - User presence status
• Secure Pipes - Stream between two users
• Speech to text - Talk to malten and write speech
• Decentralisation - Interconnect multiple Malten servers
• EventSource support - Real time messaging directly to the browser
• Client side encryption - AES-256 on the user side so we never see the data
• Configurable stream TTL - Per stream configuration to increase or decrease for DMs or Notes

Usage

Use the Go toolchain to install

go get github.com/asim/malten

Or download the latest release binary

• https://github.com/asim/malten/releases/latest

Malten is a self executable, to run it simply do

./malten

Browse to localhost:9090 or host behind an nginx proxy

OpenAI

To enable conversations using ChatGPT specify the OPENAI_API_KEY environment variable

OPENAI_API_KEY=xxxx ./malten

To invoke ChatGPT use /malten as the prefix to your command

API

Below is a high level overview of the API

Streams

To create a new stream

POST /streams

Params used

stream: string e.g stream=foo
private: bool e.g private=true

To retrieve a list of streams

GET /streams

Events

Get server sent events for real time messages

var events = new EventSource("/events")

Params for events

stream: string e.g stream=foo

Event structure

Id string
Type string e.g message or event
Created int64 e.g Unix nano timestamp
Payload object

Commands

Commands are sent to ChatGPT or other related systems that are listening

POST /commands

Send the params

stream: string e.g stream=foo
prompt: string e.g prompt=hello

The response to commands are seen within messages and events

Messages

To retrieve messages for a stream

GET /messages

Params for messages

stream: string e.g stream=foo
limit: integer e.g limit=25
direction: integer e.g direction=1 or direction=-1
last: timestamp e.g last=1652035113591000000

An example retrieval

GET /messages?direction=1&limit=25&last=1652035113591000000&stream=foo

To create messages

POST /messages

Params for messages

stream: string e.g stream=foo
message: string e.g message=helloworld

An example

POST /messages?stream=foo&message=helloworld