Deploying Citadel

Single-user, localhost-only systemd user unit. Designed to outlive gc-supervisor outages — the dashboard is exactly what Charlie wants open when gc is misbehaving, so it must not be gc-supervisor-managed.

One-time install

# 1. Build everything
cd /home/charlie/citadel
npm install
npm run build

# 2. Link the unit into the user-level systemd dir
mkdir -p ~/.config/systemd/user
cp deploy/citadel.service ~/.config/systemd/user/

# 3. Enable + start
systemctl --user daemon-reload
systemctl --user enable --now citadel.service

Browse to http://127.0.0.1:8081.

Updating

cd /home/charlie/citadel
git pull          # or your usual update mechanism
npm install
npm run build
systemctl --user restart citadel.service

Diagnostics

systemctl --user status citadel.service
journalctl --user -u citadel.service -f
ss -tln 'sport = :8081'        # port-in-use check
curl -fsS http://127.0.0.1:8081/api/health  # smoke test

Kill switch

THRIVA_ADMIN_DASHBOARD_DISABLED=1 systemctl --user start citadel.service
# → the service refuses to bind the listener; clean exit.

For permanent disable: systemctl --user disable --now citadel.service.

Notes

Bound to 127.0.0.1:8081 only (not 0.0.0.0) — see docs/SECURITY.md for the DNS-rebinding posture.
gc-supervisor outage takes the dashboard's data with it; the dashboard SHELL stays up (renders the cached/empty state) until supervisor returns.
Audit log is appended to /home/charlie/thriva-dev/.gc/events.jsonl (durable channel; survives dolt-hq corruption).

FilesExpand file tree

README.md

Latest commit

History