Skip to content

Releases: lexuss1979/ssh-commander

v0.1.1 — Prebuilt Docker images

Choose a tag to compare

@lexuss1979 lexuss1979 released this 09 Sep 23:32

Install ssh-commander on your own computer with Docker Compose, without cloning the repository or installing Node/npm. Prebuilt images support linux/amd64 and linux/arm64.

Installation and updates · Инструкция на русском · Download release Compose · GHCR package

Save the Compose file in a new directory as compose.yaml, then run docker compose up -d and open http://localhost:8080. Set an app password; you can skip the AI key and use the manual tools first.

Changes

  • Prebuilt Docker images with an exact version, localhost-only ports and persistent local data.
  • CI for server/web builds, 963 unit tests, web lint and dependency audits.
  • Docker runtime updated to Node.js 22 LTS.
  • Backup, update and source-build migration instructions, security policy, private vulnerability reporting and issue forms.
  • Two screenshots of the current interface; existing demos preserved.

No application data format changes. Before updating, stop the previous instance and back up both data/ and keys/. Never run two instances against the same data directory. Existing passwords and AI settings remain in data/settings.json; changing seed environment variables does not override them.

Verification

Successful release workflow: builds, 963 tests, lint (five existing warnings), both dependency audits, and image smoke checks on amd64 and emulated arm64. Anonymous pulls of both platforms and clean Compose installation, onboarding without AI, and data preservation after recreation were additionally verified on Windows with Docker Desktop. Physical ARM hardware and native macOS were not tested.

Source commit: bb93d661a6100f08d6971a001fe339839a110ba4 (v0.1.1). Release tooling fixes were applied in main; the existing candidate image and tagged application source were preserved.

Image: ghcr.io/lexuss1979/ssh-commander:0.1.1

Multi-platform digest:

ghcr.io/lexuss1979/ssh-commander@sha256:28866d3748649ff9d06913189d70b22d7281f29c92e172d5c74c7198e0ded676

latest points to the same digest. The source-build and hot-reload development modes remain available.

ssh-commander v0.1.0

Choose a tag to compare

@lexuss1979 lexuss1979 released this 09 Sep 22:25

ssh-commander v0.1.0

Your personal AI DevOps: get your project running on a VPS, and get help when something breaks.

This first public release lets you describe a task and have an AI agent investigate and work on your actual Linux server over SSH. You don't need to carry commands and output between a separate chat and terminal. The agent asks for approval before making changes; terminal, files and Docker tools remain available when you want to take over.

The app is free and MIT-licensed, with no ssh-commander subscription. You pay for your servers and AI API usage with your own key. This is an interactive assistant, not an autonomous 24/7 on-call service; review proposed actions and keep backups.

Highlights

  • SSH terminal tabs, command history and sessions that survive page reloads.
  • SFTP file manager with editing, directory transfers, search and live log viewing.
  • Docker Explorer with Compose v2, container terminals and live logs.
  • PostgreSQL/MySQL consoles, systemd, Nginx, cron, package updates and SSH tunnels.
  • AI plan mode, per-server memory and cost tracking with your own API key.
  • English and Russian UI, dark and light themes, and settings without a restart.

Install

Requires Git and Docker with Compose v2.

git clone https://github.com/lexuss1979/ssh-commander.git
cd ssh-commander
git checkout v0.1.0
docker compose up -d --build

Open http://localhost:8080, set a password, add your AI API key and connect your server over SSH. No .env file is required. You can skip the key to use the manual tools and add it later in Settings. Run the app on your own computer; it does not rent a VPS for you.

Demos

Animated previews are available in the README.

Before You Run It

This is a single-user, localhost-only tool, not an internet-facing server panel.
SSH/DB passwords and the AI API key are stored in plain text in the local data
volume. AI prompts and tool results go to your configured provider, which can
also be self-hosted. SSH tunnels are accessible to other local processes.

MIT-licensed, published as-is, maintained on a best-effort basis with no SLA.
Please read the README's Security section before use.