SnapOtter processes files entirely on your infrastructure. No telemetry, no external API calls, no phone-home behavior. Files never leave the container.
The container runs as a dedicated non-root user (`snapotter`) with all Linux capabilities dropped except the minimum required set. For the full vulnerability disclosure policy and security architecture, see [SECURITY.md](https://github.com/snapotter-hq/SnapOtter/blob/main/SECURITY.md) on GitHub.
## Container Hardening
The [default docker-compose.yml](https://github.com/snapotter-hq/SnapOtter/blob/main/docker/docker-compose.yml) includes production security hardening. Here is a breakdown of each option and why it matters:
```yaml
services:
SnapOtter:
image:snapotter/snapotter:latest
ports:
# Bind to localhost only for internet-facing deployments:
`security_opt: [no-new-privileges:true]` is intentionally omitted. The entrypoint starts as root to fix volume ownership, then drops to the `snapotter` user via [gosu](https://github.com/tianon/gosu), which requires setuid. Once the privilege drop completes, the process runs as `snapotter` with all capabilities except the five listed above removed.
If you use Kubernetes or Docker's `--user` flag to run as non-root directly (bypassing gosu), `no-new-privileges` is safe to enable.
### Why `read_only` Is Not Set
`read_only: true` is not set because PUID/PGID remapping writes to `/etc/passwd` and `/etc/group` at startup. If you use Docker's `--user` flag or Kubernetes `runAsUser` instead of PUID/PGID, you can safely enable a read-only root filesystem.
The only exception is **AI model downloads**: when a user installs an AI feature bundle through the UI, the container downloads model files from GitHub Releases and PyPI. These downloads happen once per bundle and are stored in the `/data` volume.
**Firewall recommendations:**
| Scenario | Outbound rule |
|---|---|
| Air-gapped (no AI) | Block all outbound traffic from the container |
| AI bundles needed | Allow HTTPS to `github.com`, `objects.githubusercontent.com`, `pypi.org`, `files.pythonhosted.org` during install, then block |
| After AI install | Block all outbound traffic -- models are cached locally |
For reverse proxy configuration (Nginx, Traefik, Caddy, Cloudflare Tunnels), see the [Deployment guide](/guide/deployment#reverse-proxy).
## Docker Secrets
For production deployments, avoid passing secrets as plain-text environment variables. The entrypoint supports Docker's `_FILE` convention: mount a secret as a file and set the corresponding `_FILE` variable to its path.
Docker Compose secrets (without Swarm) require Compose v2.23 or later.
:::
## Kubernetes Deployment
The entrypoint detects when the container is already running as non-root (e.g., via Kubernetes `runAsUser`) and skips the gosu privilege drop automatically.
**Recommended Pod SecurityContext:**
```yaml
apiVersion:apps/v1
kind:Deployment
metadata:
name:snapotter
spec:
replicas:1
selector:
matchLabels:
app:snapotter
template:
metadata:
labels:
app:snapotter
spec:
securityContext:
runAsNonRoot:true
runAsUser:999
runAsGroup:999
fsGroup:999
containers:
- name:snapotter
image:snapotter/snapotter:latest
ports:
- containerPort:1349
securityContext:
allowPrivilegeEscalation:false
capabilities:
drop:[ALL]
resources:
requests:
cpu:"1"
memory:2Gi
limits:
cpu:"4"
memory:6Gi
livenessProbe:
httpGet:
path:/api/v1/health
port:1349
initialDelaySeconds:60
periodSeconds:30
timeoutSeconds:5
readinessProbe:
httpGet:
path:/api/v1/health
port:1349
initialDelaySeconds:10
periodSeconds:10
timeoutSeconds:5
volumeMounts:
- name:data
mountPath:/data
- name:workspace
mountPath:/tmp/workspace
volumes:
- name:data
persistentVolumeClaim:
claimName:snapotter-data
- name:workspace
emptyDir:
medium:Memory
sizeLimit:2Gi
```
Since `runAsUser: 999` is set at the pod level, the entrypoint skips gosu entirely. This allows `allowPrivilegeEscalation: false` and `drop: [ALL]` capabilities without conflict.
For resource sizing, see [Hardware Requirements](/guide/deployment#hardware-requirements).
Alternatively, stop the stack and snapshot the `SnapOtter-pgdata` volume:
```bash
docker compose down
docker run --rm -v SnapOtter-pgdata:/data -v $(pwd)/backup:/backup \
alpine tar czf /backup/snapotter-pgdata.tar.gz -C /data .
```
### User files backup
```bash
# Snapshot the app data volume (excluding re-downloadable AI models)
docker run --rm -v SnapOtter-data:/data -v $(pwd)/backup:/backup \
alpine tar czf /backup/snapotter-files.tar.gz \
--exclude='ai' --exclude='venv' -C /data .
```
AI models total up to 14 GB across all bundles. Since they are re-downloadable, exclude `/data/ai/` and `/data/venv/` from backups to save space. Only the database and user files are critical.
Download the SBOM from the release and scan it with your preferred tool:
```bash
# Scan with Grype using the CycloneDX SBOM
grype sbom:snapotter-v1.17.2-sbom.cdx.json
# Scan with Trivy using the SPDX SBOM
trivy sbom snapotter-v1.17.2-sbom.spdx.json
# Scan the Docker image directly
trivy image snapotter/snapotter:1.17.2
```
::: info
The SBOM and vulnerability scan reflect the exact image published for that release. AI model bundles installed after deployment are not included in the SBOM since they are downloaded at runtime.
