I want to self-host Linkbeam — open-source music smartlinks on Cloudflare Workers — for my label. Analyze my project and recommend the best integration approach.

## My Project Context

[Describe your stack: Cloudflare account ready? Custom domain? Meta Pixel + CAPI credentials? Existing music-shortlink deploy? Bun/Node installed?]

## Two Integration Approaches

Linkbeam is a full-stack Worker app (Astro 7 + D1 + KV + R2 + Queues), not an SDK you npm-install.

### Approach 1: Fresh self-host (RECOMMENDED)

**Best for:** New operators with their own Cloudflare account and domain.

**How it works:**
1. Clone https://github.com/DerpcatMusic/linkbeam
2. Create Cloudflare resources: D1, KV (`LINK_CACHE`), R2 (`linkbeam-artwork`), Queues (`linkbeam-conversions` + DLQ)
3. Fill placeholder IDs in `wrangler.jsonc`; set `PUBLIC_BASE_URL` to your domain
4. Run `bun run db:migrate:local` then `bun run db:migrate`
5. Set secrets: `META_PIXEL_ID`, `META_ACCESS_TOKEN` (optional: Spotify, Cloudflare Access, admin password)
6. `bun run dev` locally, then `bun run deploy`
7. Open `/admin/onboarding` — wizard validates bindings and generates wrangler snippets

**Fan flow:** Meta ad → `/:slug` fan page → `/out/:slug/:platform` redirect with server-side CAPI.

### Approach 2: Deployment fork (existing Cloudflare resources)

**Best for:** Operators who already deployed under the `music-shortlink` worker name and want to keep existing D1/KV/R2/queue IDs.

**How it works:**
1. Use https://github.com/DerpcatMusic/music-shortlink as a deployment fork
2. Develop features in the main `linkbeam` repo
3. Run `./scripts/sync-from-linkbeam.sh` to pull product code while preserving `wrangler.jsonc` and domain config
4. Deploy to your existing `links.*` domain

## Documentation (read before writing code)

- Index: https://derpcatmusic.github.io/linkbeam/llms.txt
- Full bundle: https://derpcatmusic.github.io/linkbeam/llms-full.txt
- Getting started: https://derpcatmusic.github.io/linkbeam/getting-started.html
- Configuration: https://derpcatmusic.github.io/linkbeam/configuration.html
- Admin workflow: https://derpcatmusic.github.io/linkbeam/admin.html
- Meta tracking: https://derpcatmusic.github.io/linkbeam/tracking.html

## Stack Reference

- **Runtime:** Cloudflare Workers (`src/worker.ts` — fetch + queue consumer + cron CAPI retry)
- **Framework:** Astro 7 server output via `@astrojs/cloudflare`
- **Data:** D1 (links, metrics, settings), KV (published link cache), R2 (artwork)
- **Tracking:** `src/lib/tracking.ts` — first-party `_fbp`/`_fbc` cookies, Meta CAPI via Queue, bot filtering
- **Admin:** `/admin/*` — Cloudflare Access JWT or password session
- **Routes:** `/:slug` (fan page), `/out/:slug/:platform` (click + CAPI), `/api/*` (import, subscribe)

## What I need you to do

1. Fetch https://derpcatmusic.github.io/linkbeam/llms-full.txt (or the doc pages above) for grounded context
2. Recommend Approach 1 or 2 based on my project context
3. Walk me through `wrangler.jsonc` setup with my Cloudflare resource IDs
4. Configure Meta CAPI secrets and validate with the onboarding pixel test
5. Help me create and publish my first smartlink from a Spotify or TooLost URL
6. Do not add client frameworks to fan pages — Linkbeam is server-rendered HTML by design

## Constraints

- Fan pages (`/:slug`) must stay fast: no React/Vue, minimal client JS
- Meta events must use server-side CAPI with `event_id` deduplication — do not replace with browser-only pixel
- Bot traffic is excluded from metrics — preserve `src/lib/bots.ts` behavior
- `WORKER_NAME` in wrangler vars controls onboarding resource naming snippets