deploy: replace stub docker-compose with working VPS stack

[replicas-connector]

Summary

• deploy/docker-compose.yml on dist was a one-line stub: file:///tmp/b2_freescout_compose.txt. It cannot deploy anything.
• Replace it with a real, production-shaped FreeScout + MariaDB compose stack so Nomad can stand up FreeScout on a fresh VPS.
• Add deploy/.env.example (passwords, APP_URL, SMTP) and deploy/README.md with a 4-step bring-up.

What's in the stack

• App: tiredofit/freescout (tag pinned via FREESCOUT_IMAGE_TAG), bound to APP_PORT for an upstream reverse proxy (Caddy/nginx/Traefik) to terminate TLS.
• DB: mariadb:10.11 with a healthcheck; app waits for service_healthy before starting.
• Persistence: named volumes freescout_data, freescout_logs, freescout_db.
• Config: every secret and URL is sourced from deploy/.env — nothing baked into the compose file.

Test plan

• docker compose -f deploy/docker-compose.yml --env-file deploy/.env.example config -q validates cleanly (no warnings).
• On the target VPS: cp deploy/.env.example deploy/.env, fill in real values, docker compose ... up -d, then complete the FreeScout web installer at APP_URL.
• Confirm reverse proxy forwards :443 → APP_PORT.
• Send a test outbound email via Manage → System → Mail Settings to verify SMTP creds.

---

_{Created by Jaden (jadengarza@pm.me) with Replicas
Workspace  ·  Slack Thread}

[capy-ai]

Capy auto-review is paused for this organization because the monthly auto-review limit has been reached. Increase the limit or turn it off in billing settings to resume automatic reviews.

[replicas-connector]

Pushed 4e66259 to fix several latent problems that would've made this stack fail to actually serve FreeScout once docker compose up returned:

Why the original stack didn't work

The tiredofit/freescout image was rebranded to nfrastack/freescout in May 2026. The DockerHub tag still publishes, but it's the new image — and the new image broke compatibility with the env var names and volume layout this compose was built around:

What we had	Why it broke	Fix
freescout_data:/www/html only	State (sessions, cache, installed modules, generated .env) lives at /data in the new image; /www/html is now optional and gets clobbered on every container start by ENABLE_AUTO_UPDATE	Mount freescout_data:/data (+ freescout_logs:/logs)
ENABLE_SMTP / SMTP_HOST / SMTP_PORT / SMTP_USER / SMTP_PASS / SMTP_FROM_ADDRESS	Removed in the new image — mail vars must use the FREESCOUT_ passthrough prefix	Switch to FREESCOUT_MAIL_HOST, FREESCOUT_MAIL_PORT, FREESCOUT_MAIL_USERNAME, FREESCOUT_MAIL_PASSWORD, FREESCOUT_MAIL_FROM_ADDRESS, FREESCOUT_MAIL_FROM_NAME, FREESCOUT_MAIL_ENCRYPTION
SITE_URL	Legacy alias; canonical is now APP_URL	Use APP_URL
DISPLAY_ERRORS, CONTAINER_ENABLE_MONITORING, VIRTUAL_HOST	Dead vars on the new image	Removed
healthcheck: curl -fsS http://localhost/	The new alpine-based image doesn't ship curl by default — the healthcheck would always report unhealthy and the depends_on: service_healthy chain would stall	Use a bash TCP probe (exec 3<>/dev/tcp/127.0.0.1/80) and lift start_period to 120s for first-boot schema migration
(missing) DB_TYPE	Defaults to mysql; with a MariaDB server attached this still works but the entrypoint logs warnings and the wrong driver lands in .env	Set DB_TYPE=mariadb

Also:

• Updated deploy/.env.example to match the new var names + added SMTP_ENCRYPTION (defaults to tls).
• Updated deploy/README.md to point at the canonical image + document the FREESCOUT_* prefix for adding any other FreeScout .env setting.

docker compose -f deploy/docker-compose.yml --env-file deploy/.env.example config -q still validates cleanly.

The unrelated Continuous AI: New Dev Agent check on this PR failed on the previous commit with no diagnostics surfaced through the GitHub API — it doesn't appear to gate anything.