Prototype: deploy loopover-miner-ui as a demo Cloudflare Worker (informs #5229's web-dashboard reuse decision)

[JSONbored]

Problem

apps/loopover-miner-ui is a real, working self-host operator dashboard (React + TanStack Router, with local API plugins for auth/run-state/portfolio-queue/ledgers/governor/ranked-candidates), but today it only exists as something an individual AMS miner builds and runs locally via a systemd service (systemd/loopover-miner-ui.service.example) — deliberately excluded from this repo's own CI build/deploy pipeline (ci.yml's "UI build" step and ui-deploy.yml both explicitly skip it; test/unit/ci-ui-build-openapi.test.ts asserts this). That means there is currently no way to show anyone what AMS's dashboard looks like without them self-hosting it first, and no CI-validated build step to hang anything — like bundle-size tracking — off of.

Meanwhile #5229 ("Spec: client-surface strategy for hosted AMS") is an open decision issue asking whether apps/loopover-miner-ui can become hosted AMS's own web dashboard via a "hosted-mode config flag," reusing the self-host codebase rather than forking it (the "identical core + thin additive wrapper" pattern this whole milestone recommends). That decision is currently being made in the abstract, with no working prototype to point at.

Area

AMS / Cloud architecture, client-surface

Proposal

Deploy the existing apps/loopover-miner-ui codebase as its own Cloudflare Worker on a subdomain, mirroring the pattern apps/loopover-ui already uses (.github/workflows/ui-deploy.yml + ci.yml's "UI build" step) — but running against a mocked/synthetic data layer instead of a real backend, since none of the actual hosted-AMS infrastructure (auth, multi-tenancy, shared datastore — #4938, #4940, #4941) needs to exist yet for a demo. Concretely: swap the local API plugins (vite-auth.ts, vite-run-state-api.ts, vite-portfolio-queue-api.ts, etc.) for a build-time-flagged mock layer that serves realistic sample data, and add a dedicated CI build+deploy step for this app the same way loopover-ui already has one.

This is deliberately scoped as a prototype/demo, not a commitment to the hosted-AMS architecture:

• It gives sales/marketing/roadmap conversations something real to point at before any hosted backend exists.
• It's a concrete, working data point for Spec: client-surface strategy for hosted AMS (web dashboard vs. mobile/desktop apps vs. CLI-only) #5229's own open question ("can we reuse the self-host dashboard for hosted") rather than a purely theoretical evaluation.
• Once this app has a real CI-validated build step, Codecov Bundle Analysis (already wired up for loopover-ui in feat(ci): upload loopover-ui bundle stats to Codecov #5961) becomes viable for it too — closing the exact gap that surfaced while building feat(ci): upload loopover-ui bundle stats to Codecov #5961: bundle analysis needs a CI build step to hook into, and loopover-miner-ui currently has none.

Deliverables

• A apps/loopover-miner-ui "demo mode" (mocked data behind a build-time flag, e.g. VITE_DEMO_MODE) that runs with zero real backend and no operator credentials.
• A dedicated Cloudflare Worker + subdomain deploy for the demo build, mirroring ui-deploy.yml's structure. Start behind manual workflow_dispatch (matching how ui-deploy.yml itself already works today) rather than wiring it into every PR — a stale or broken demo redeploying on every commit is worse than a manually-triggered one.
• Once the build step exists, extend Codecov Bundle Analysis to cover loopover-miner-ui too (companion to feat(ci): upload loopover-ui bundle stats to Codecov #5961).

Explicitly NOT in scope for this issue

• Any real multi-tenant hosted AMS backend — auth/identity, billing, tenant isolation, shared datastore, secrets-at-scale. All separately tracked (Design a client/server split for AMS #4938, Add tenant-scoping columns to all local ledger schemas #4939, Design: storage-abstraction interface for AMS's node:sqlite stores (D1/Postgres seam, Postgres-lead) #4940, Design an auth/identity layer for AMS #4941, Design: provisioning API/CLI to stand up a new hosted AMS tenant instance #5220–Draft: SLA and incident-response commitment for hosted AMS #5227).
• A final decision on whether hosted AMS ships a web dashboard at all, or in what form — that's Spec: client-surface strategy for hosted AMS (web dashboard vs. mobile/desktop apps vs. CLI-only) #5229's call. This issue's output (a working demo) is meant to inform that decision with a concrete prototype, not preempt it.

Boundaries

• No real operator credentials, GitHub tokens, or any actual miner's local ledger/governor data in the demo build — mocked/synthetic data only, reviewed for anything that could look like a real trust score, reward value, or private scoring detail before it ships anywhere public.
• Do not treat shipping this demo as "hosted AMS is now decided" — Spec: client-surface strategy for hosted AMS (web dashboard vs. mobile/desktop apps vs. CLI-only) #5229 remains the actual decision point.

Resources

• apps/loopover-miner-ui (existing self-host dashboard)
• .github/workflows/ui-deploy.yml + ci.yml's "UI build" step (the pattern to mirror; also see test/unit/ci-ui-build-openapi.test.ts for why loopover-miner-ui is currently excluded)
• Spec: client-surface strategy for hosted AMS (web dashboard vs. mobile/desktop apps vs. CLI-only) #5229 (the client-surface decision this prototype informs)
• feat(ci): upload loopover-ui bundle stats to Codecov #5961 (the loopover-ui bundle-analysis PR that surfaced this gap)
• Epic Epic: AMS Cloud Readiness #5230 / milestone "AMS Cloud Readiness"

[linear-code]

JSO-4114

[JSONbored]

Releasing from maintainer-only — this milestone is genuinely gated on the undecided 'should we build hosted, and when' question, but this specific issue isn't. Explicitly marked in its own body as 'deliberately scoped as a prototype/demo, not a commitment to the hosted-AMS architecture' — mirrors the existing loopover-ui deploy precedent, uses only mocked/synthetic data, no real backend or multi-tenancy, explicitly informs but does not preempt #5229's decision.

[RealDiligent]

Working on this — will open a PR with demo-mode mock data + workflow_dispatch CF Worker deploy mirroring ui-deploy.yml, plus Codecov bundle analysis for loopover-miner-ui.