retire-ledger

A dividend/retirement-income ledger tool (retire-ledger.silvarion.org). A single hand-authored html/index.html — no build step, no framework — deployed to Cloudflare Workers as static assets.

See AGENT.md for the guiding principles this project follows.

History

This was previously deployed via a docker-compose stack (a small Python companion server, ledger-server.py, persisting the saved fund-table config server-side to a bind-mounted config.json) under infrastructure/homelab/containers/docker-compose/retire-ledger. That stack is retired in favor of this repo deployed straight to Cloudflare Workers — a static deploy has no persistent container to run that companion server in. Concretely, that meant:

• Config persistence moved from server-side to this browser's localStorage (see saveConfigToLocalStorage()/ loadConfigFromLocalStorage() in html/index.html). The saved fund table no longer syncs across devices/browsers — each browser keeps its own copy. Use the existing Download/Upload buttons to carry a config from one browser/device to another.
• The Eulerpool API key was already kept in-memory only (never sent to this app's own server, never persisted) — that behavior is unchanged by this migration; it's what let this become a pure static site with no backend at all despite calling a keyed third-party API client-side. (This has since changed by explicit request, 2026-09-18: API keys — now Eulerpool and Twelve Data both — persist to this browser's own localStorage when you click "Save as new defaults", kept in a separate storage key from the shareable config so they never end up in a Download/Upload file. See AGENT.md's "API keys" section for the current behavior and its tradeoff.)

The old docker-compose directory has not been deleted — it's left in place for reference/rollback. The cutover plan (all manual, outside this repo/this session — no access to any of these hosts):

1. Deploy this repo to Cloudflare (see below).
2. Re-point retire-ledger.silvarion.org's DNS/routing at the new Cloudflare deployment (see "Custom domain / DNS cutover" below) and confirm it actually serves the new site.
3. Only then decommission the docker-compose stack: remove the retire-ledger entry from dockge_stacks in infrastructure/homelab/proxmox/srv-proxmox01/base/lxc/dockge/core's group_vars, redeploy that host, then remove the old stack directory — and update the Uptime Kuma monitor/Prometheus blackbox target for retire-ledger.silvarion.org to the {inf_tech: cloudflare-workers, inf_category: external} labels resume-silvarion's own entry already uses, since the target is no longer a dockge-fronted service.

Requirements

None beyond a browser. No build step — html/index.html is served as-is.

Local preview

npx serve html
# or: python3 -m http.server 8000 -d html

Deploying to Cloudflare (Workers Builds)

1. Push this repo to a Git provider Cloudflare can read from.
2. In the Cloudflare dashboard: Workers & Pages → Create → Connect to Git, pick this repo.
3. Build settings (Settings → Build on the created Worker):
   • Build command: (leave empty — no build step)
   • Deploy command: npx wrangler deploy (the default)
   • Version command: npx wrangler versions upload (the default — used for preview builds on non-production branches)
4. wrangler.toml defines an assets-only Worker ([assets] directory = "html", no Worker script) — Cloudflare serves html/index.html directly.

Direct upload with Wrangler (no Git integration)

npx wrangler deploy

Custom domain / DNS cutover

retire-ledger.silvarion.org currently reaches the docker-compose stack via opnsense-caddy (the internet-facing OPNsense firewall's own Caddy instance, 10.48.10.250) forwarding straight to the container — this routing is not managed by the silvarion.webservers.caddy role or any repo in this workspace (confirmed: it has no retire-ledger entry in infrastructure/homelab/proxmox/srv-proxmox01/base/lxc/caddy/core's caddy_sites), so it can't be changed from here.

To cut over:

1. In the Cloudflare dashboard, on this Worker: Settings → Domains & Routes → Add → Custom Domain → retire-ledger.silvarion.org. Since the silvarion.org zone is already on Cloudflare, this creates/updates the DNS record to point at the Worker automatically.
2. That DNS change alone doesn't remove the existing opnsense-caddy forwarding rule for this hostname — whichever one Cloudflare/DNS resolution actually routes to first wins. Remove or repoint that rule on the OPNsense box once the Custom Domain is confirmed live, so traffic isn't split or flapping between the old and new deployments.
3. Verify with curl -sI https://retire-ledger.silvarion.org — a server: cloudflare response with no via: ... Caddy header means it's hitting the Worker directly, not the old chain.

Project structure

html/index.html   The entire app — markup, inline CSS, inline JS
html/_headers     Security headers / CSP for Cloudflare
wrangler.toml     Cloudflare Workers Builds config (assets-only, no Worker script)