Vexillon

Install

Up and running
in minutes.

Install MIRA natively on Linux, macOS, Windows or WSL — or run the bundled Docker image. However you start her, a guided setup walks you (and your family) through the rest.

Linux macOS Windows WSL Docker
Your platform of choiceRuns natively on Linux, macOS, Windows and WSL — or as a Docker container. Use whatever you already have.
A machine at homeA home server, mini PC, or spare laptop. MIRA is lightweight by design.
An AI modelAn API key (Anthropic, OpenAI, Gemini, OpenRouter, DeepSeek) — or a local LLM, for fully offline.
1

Install her

The quickest path is Docker — clone and bring the stack up. The compose file mounts ./data, the one folder that holds your whole install. Prefer a native install? Linux, macOS, Windows and WSL each have a guide in the repo.

docker
$ git clone https://github.com/Vexillon-ai/MIRA.git
$ cd MIRA
$ docker compose up -d
✓ MIRA is awake at http://localhost:8080
2

Let the guided setup take over

Open the web UI and MIRA's onboarding walks you through the rest — creating your admin account, choosing an AI model (bring your own key, or point her at a local model for fully offline), and connecting your first channel. No config files to hand-edit.

3

Bring in the rest of the family

Add more channels anytime under Settings → Channel Accounts — a Telegram or Discord token, a Signal number, an email account. Each person can link their own, or share one household bot.

4

Back it up

Everything lives under the data directory. Use the one-click (optionally encrypted) backup in Settings → Server, or just copy ./data. Need a different port?

bash
$ MIRA_PORT=9090 docker compose up -d

Prefer to build from source?

MIRA is written in Rust (edition 2024, Rust 1.85+). Clone the repo and cargo build --release. The full developer setup, channel guides and architecture docs live in the repository.

Read the docs on GitHub

Welcome her in.

Free, open source, and yours. The whole household, one quiet helper.

Star on GitHub Star on GitHub