Selfhost AI CLI
An installer deploys selected AI services on a separate VPS and provides commands to configure and maintain the stack
- 955 ★ on the repository
- Last commit
- Apache-2.0
Broad access
Why this level
- The sudo installer changes system packages, networking and firewall settings
- The `.env` file and Welcome page contain selected services’ secrets
- Updating can replace infrastructure customizations and restart running services
Install
- Works in
Manual install
git clone https://github.com/kossakovsky/selfhost-ai && cd selfhost-ai && sudo bash ./scripts/install.shUse a prepared separate VPS. Review the script, domain and resources before running with administrator privileges.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The sudo installer changes system packages, networking and firewall settings
- The `.env` file and Welcome page contain selected services’ secrets
- Updating can replace infrastructure customizations and restart running services
This is third-party code. Review the repository files before installing.
What it does
Selfhost AI installs a selection of AI-project services on your own server. An interactive wizard lets you choose components, prepares the host and starts Docker Compose. CLI commands maintain the resulting stack. Components include Ollama and Open WebUI for model use. You can choose n8n for automation and separate storage and RAG services for document retrieval. Optional n8n-MCP exposes node documentation and n8n workflow tools to coding agents. The installer creates service subdomains and a page containing access details. After installation, `make status` shows containers and `make doctor` checks infrastructure. Each deployed service retains its own configuration and terms.
Who it is for. For developers and administrators prepared to maintain a separate VPS with selected AI services and agent integrations.
Good fit when
- You need several compatible AI services on a separate server
- You want shared diagnostic and update commands for a Docker stack
- A team needs its own n8n alongside selected AI components
- You need the server environment ready before connecting a coding agent to n8n-MCP
Not a fit when
- You need one small MCP without server setup
- The installer cannot have administrator privileges on the host
- You only have a home computer or shared hosting without a public VPS
Example request
Plan Selfhost AI installation on a new VPS with only n8n and its required dependencies. Check resources, DNS and host changes before running the script.Limitations
The supported route needs a separate public-IP VPS with 64-bit Ubuntu 24.04 LTS and a wildcard domain A record pointing to it. A small subset needs 4 GB RAM, 2 cores and 40 GB disk; all services need at least 20 GB RAM, 4 cores and 60 GB, before additional model/data storage. The installer sends usage events through Scarf; SCARF_ANALYTICS=false in .env disables its telemetry. A local model does not eliminate other components’ external calls; inspect connections in the chosen stack.
How to disable. Run `make stop` in the installation directory to stop the stack. Remove service connections from agents and revoke their keys separately. Preserve container data and `.env` before removal. Stopping Compose does not restore system or firewall settings changed by the installer.
FAQ
How do I start the installer?
Clone the repository and run `sudo bash ./scripts/install.sh` from its directory. It changes the host, so read it first and prepare a separate server.
What changes on the server?
The installer updates the system, configures firewall and login protection, installs Docker and starts selected services. Its scope extends beyond one container.
Where are service addresses and passwords?
On `welcome.<your-domain>`, printed at the end of installation. It contains service credentials; restrict access and do not publish screenshots containing passwords.
Does updating preserve manual file changes?
Not all of them. `make update` resets tracked files including `Caddyfile` and `docker-compose.yml`. The author recommends `caddy-addon/site-*.conf` and `docker-compose.override.yml` for customizations.
Must I import bundled n8n workflows?
No. Import is offered separately when selecting n8n. Review needed workflows and their connections first; a third-party template can call external services.
Which licenses and costs apply?
The installer uses Apache-2.0; components retain their own licenses. VPS hosting, the domain and external APIs have separate costs.
Discussion
No questions or reviews yet.
Related
CLI toolsA Go CLI that compares a server with its last snapshot, reports only changes that matter and serves them to an agent via MCP
A skill and a Go binary: with your approval the agent installs apps from a 93-app catalog on your home server via Docker Compose
3X-UI on your VPS
3x-ui-vps
Skill that deploys the 3X-UI panel and Xray VLESS behind nginx on an Ubuntu server and maintains it through bundled scripts
1C AI sandbox client-server
1c-ai-sandbox-client-server
A devcontainer plus a Hyper-V VM with a 1C server that isolate an AI agent from the host machine so a wrong command cannot touch real files or drives
Container by Aerovato
Container
A CLI for persistent per-project Linux workspaces in Docker or Podman that keep the agent away from the rest of your system
An AI assistant for Kubernetes that turns plain-language requests into kubectl operations and doubles as an MCP server
README in short
The wizard generates secrets for selected services in `.env`. The user keeps them locally; Make commands manage the assembled stack.