Install with the script

install.sh sets up a complete DocMyRepos with Docker on your own machine: it checks that Docker works, asks four questions, generates every secret, writes a compose file and an .env, starts the app and its Postgres database, and waits until the app is healthy. It does not register your GitHub App: you enter that in the app after the first sign-in, where it is validated and stored encrypted.

The other way to install is the clone-and-compose path in Getting started.

Run it

You need Linux, macOS or WSL, curl, and Docker with Compose v2 running.

curl -fsSL https://github.com/Dash-AI-Tools/doc-my-repos/releases/download/v0.2.0/install.sh | sh

The URL is the install.sh asset on the GitHub Release for that tag. Replace v0.2.0 with the release you want; the script installs the container image of its own version (ghcr.io/dash-ai-tools/doc-my-repos:0.2.0). To pass flags to a piped script, put them after sh -s --:

curl -fsSL https://github.com/Dash-AI-Tools/doc-my-repos/releases/download/v0.2.0/install.sh | sh -s -- --dry-run

Read it first. It is one file of about 700 lines. Fetch it, read it, then run it from disk:

curl -fsSLO https://github.com/Dash-AI-Tools/doc-my-repos/releases/download/v0.2.0/install.sh
less install.sh
sh install.sh

The script contacts only the container registry (to pull the image) and http://127.0.0.1 (the health check). It sends no telemetry and fetches nothing else.

What it asks

Each question shows its default in brackets. Press Enter to accept it.

Question	Default	Notes
Install folder	./docmyrepos	Created if missing. It must not already hold a .env or docker-compose.yml.
Port for the portal	3000	Asked again if the port is busy.
First administrator email	none	Used to sign in the first time.
Administrator password	generated	Typed without echo, 12 or more characters, no ' or \. Press Enter to generate one.

It then shows a summary and asks for confirmation before it writes anything. A generated password is shown once, at the end, and is also in the .env file.

What it writes and starts

Two files in the install folder, both readable only by you (the folder is 0700, .env is 0600):

• docker-compose.yml: the app and Postgres 17, with the same hardening as the repository's compose file (read-only root file system, no capabilities, no new privileges, a process limit, rotated logs). The app is published on 127.0.0.1 only, and the database is not published at all.
• .env: the generated secrets (SETTINGS_ENCRYPTION_KEY, SESSION_SECRET, SYNC_SECRET, the database password), the image, the port, and the first administrator.

Each install gets its own Compose project name (docmyrepos- and six random hex digits), so two installs never share a database volume.

Then it pulls the image (unless it is already on the machine), starts the containers, and waits for the health check. Files are written completely or not at all: if something fails before the files are in place, the ones it created are removed. It never overwrites an existing installation.

After it finishes

1. Open the address it prints and sign in as the administrator.
2. Go to Admin > Workspace configuration and enter your GitHub App (GitHub App setup explains how to register one).
3. On the Repository access tab, press Run first sync.
4. Remove ADMIN_EMAIL and ADMIN_PASSWORD from .env, then run docker compose up -d in the install folder.

Back up the .env file with your database. SETTINGS_ENCRYPTION_KEY decrypts the secrets stored in the database; without it they cannot be read and must be entered again.

Flags and variables

Every answer can be given up front. With --yes, or when there is no terminal, nothing is asked: a missing answer is listed and the script exits with status 2 before writing anything.

Flag	Variable	Meaning
--dir <folder>	DOCMYREPOS_DIR	Install folder.
--port <number>	DOCMYREPOS_PORT	Port on this machine.
--admin-email <e>	DOCMYREPOS_ADMIN_EMAIL	First administrator's email.
none	DOCMYREPOS_ADMIN_PASSWORD	Administrator password. Environment only, so it never reaches ps.
--image <reference>	DOCMYREPOS_IMAGE	Container image, for a mirror or a build of your own.
--upgrade	none	Upgrade the installation in --dir to the image (see Upgrade).
--yes	none	Never ask.
--dry-run	none	Show the checks, the plan and the files, and write and start nothing.
-h, --help	none	Show the options.

An example for automation:

DOCMYREPOS_ADMIN_PASSWORD='a long password' \
  sh install.sh --yes --dir /opt/docmyrepos --port 8080 --admin-email you@example.com

Without a given password, --yes generates one and prints it once.

Upgrade

Run the newer release's script with --upgrade and the folder of your installation. The script's own version is the target, so using the script of release v0.2.0 upgrades to image 0.2.0:

curl -fsSL https://github.com/Dash-AI-Tools/doc-my-repos/releases/download/v0.2.0/install.sh | sh -s -- --upgrade --dir ./docmyrepos

Add --dry-run to see what it would do, or --image <reference> to upgrade to a specific image (a mirror, a build of your own). In a terminal it shows the plan and asks before it starts; with --yes, or with no terminal, it does not ask.

What it does, in this order:

1. Checks that the folder holds an installation made by the script, and that the installation is not already on the target version (then it says so and stops, changing nothing).
2. Refuses a downgrade: when both versions are known (MAJOR.MINOR.PATCH tags of the same image) and the target is older, it stops, because database migrations only go forward.
3. Pulls the new image. If that fails, nothing has changed.
4. Saves a backup of the database to backup-<UTC time>.sql in the install folder, readable only by you. If the database is not running or the backup is empty or fails, it stops and nothing has changed.
5. Changes only the DOCMYREPOS_IMAGE line in .env. The compose file, every other setting and your data are kept.
6. Restarts and waits for the health check.

It never rolls back by itself, because the new version may already have migrated the database. If the new version does not become healthy, it prints the end of the app log, the backup file and the commands to go back:

cd ./docmyrepos
docker compose stop app
docker compose exec -T db psql -U docmyrepos -d docmyrepos < backup-<time>.sql
# in .env, set DOCMYREPOS_IMAGE back to the previous value
docker compose up -d --wait

The backup is a plain SQL dump that drops and recreates the tables when restored. It holds your users and stored settings, so treat it like .env. Old backups are never deleted for you: remove them when you are sure the new version works.

The compose file is not rewritten. If a release changes it, its notes say so; to compare, run a fresh install of the new release into another folder and diff the two docker-compose.yml files.

Stop and remove

• Stop: docker compose stop in the install folder.
• Remove everything, including the data: docker compose down -v deletes the database volume. Keep a backup if the comments or users matter.

If it stops

Message	What to do
Docker is not installed / not reachable	Install Docker, or start it (Docker Desktop, or sudo systemctl start docker).
Docker Compose v2 is not installed	Install the Compose plugin; the old docker-compose (v1) is not supported.
Port is already in use	Choose another port with --port.
Already contains an installation	Choose another folder. The script never changes an existing one.
Could not pull the image	Check the network and the image name; a mirror goes in --image.
Could not reach a healthy state	Read the log it prints, or run docker compose logs app in the install folder.

When start-up fails the files stay in place so you can read the logs and fix the cause.