SITES HANDOFF — bestdadjokes.lol, helpme.tips, raspiinthesky.com

Audience: whoever continues the sites + monetization track. Infrastructure handoff is in HANDOFF.md (same repo) — this doc covers the three sites only. Written by the ZCode agent that built them. Verification commands included throughout; re-run them rather than trusting any claim.

1. Shared platform (read first)

All three sites are static builds served by the same edge:

Generator pattern: each site has build.py (Python 3.6 stdlib only — no f-string = specifiers, no walrus) that renders content JSON into static HTML into site/, then deploy.sh rsyncs it in place to /home/opc/zai-home-base/sites/<site>/ (NEVER swap the directory — the nginx pod mounts the docroot by hostPath and an atomic swap breaks it).
Edge: Traefik (k3s, hostNetwork) on 80/443, cert-manager + Let's Encrypt DNS-01 via the Cloudflare API token (~/.cloudflare-token, 600 perms). Cloudflare proxies all domains (orange cloud).
Ad integration: slots are marked  in the build templates. Nothing is wired — the user's GAM wrapper spec was never delivered. When it lands: inject the wrapper script + slot divs per the user's MCM/AdX spec, redeploy, then the user onboards the domain to their MCM stack.
Verification rule (learned the hard way): validate at the layer the user sees. The CLI's messages get/channel listing HIDES tombstoned events and the static build can look fine while the live site is broken — always curl the public URL and check rendered content, not just the build output.

Per-site deploys:

cd /home/opc/zai-home-base/bestdadjokes && ./deploy.sh   # wait: dir is bdj/
cd /home/opc/zai-home-base/bdj && ./deploy.sh            # bestdadjokes.lol
cd /home/opc/zai-home-base/helpme-tips && ./deploy.sh    # helpme.tips

2. bestdadjokes.lol — fully operational, zero revenue so far

Live: https://bestdadjokes.lol (plus welcome-everyone channel on Buzz for its audience thread). Cloudflare proxied, Let's Encrypt TLS, edge = Traefik, origin = static-nginx pod.

Content (as of handoff):

322 jokes in bdj/content/jokes.json — APPEND-ONLY (ids = array index; the web client's share feature and the generator reference ids)
14 category pages + 9 keyword topic pages + generator + joke-of-the-day (with RSS) + search + about/contact/privacy/terms (privacy has the Google ad-cookie language required for AdSense/MCM review)
Changelog: every content/schema change is a git commit; subagent QA passes ran 3 times (full crawl + rendered-content checks)

Known issues (open):

None open. Historic: two missing-stylesheet bugs (build template lacked the — caught by the user and by a QA subagent), a broken pager link, a DB-write-no-op that silently skipped a purge (see HANDOFF.md §6).

Monetization status: NOTHING LIVE. Strategy (agreed with the user):

Organic SEO baseline (currently zero traffic — domain is fresh)
Social distribution — BLOCKED on user providing social accounts
MCM/AdX onboarding — BLOCKED on user's GAM service account + wrapper spec
Funded traffic tests — BLOCKED on traffic-buying account + budget The ad slots are in the templates; integration is a config paste when the wrapper spec arrives.

Pending work queue:

Content expansion (+10 jokes/week cadence was planned, not ongoing)
Topic-page intros enrichment (9 pages have thin intros)
GSC submission (blocked on user creating the property)
dadjokes.guru sister build (design only)

3. helpme.tips — live, awaiting ad integration

Live: https://helpme.tips (Cloudflare proxied, TLS, same edge). Cloudflare DNS records created via API; the zone is in the CF token scope.

Content:

8 articles × 3 pages = 24 pages (kitchen/sleep/coffee/habits/home/tech/ groceries/fitness), about/contact/privacy/terms with Google ad language
Ad slots marked  in the build template
Deploy: cd /home/opc/zai-home-base/helpme-tips && ./deploy.sh (build → in-place rsync → origin smoke checks)

Known issues: the QA subagent fixed a missing stylesheet link and a pager bug pre-launch; full 30-page crawl passed after. Nothing open.

Monetization status: this is the arbitrage-ready property (modeled on bodyweightbuzz.com — see the arbitrage-play memory). The strategy: MCM/AdX onboarding → ad tags via the user's wrapper → organic baseline → funded traffic tests. BLOCKED on the same user unlocks as bestdadjokes.

Pending:

Ad integration when the wrapper spec arrives
Content batch #2 (+10 articles) — planned, not started

4. raspiinthesky.com — NOT STARTED

Domain owned, in Cloudflare (zone in the token scope — verify with the zones API if in doubt). No content, no deploy, no design. Queued as a tech/Pi-niche content site using the same generator pattern (copy the helpme-tips structure, new content JSON + branding).

5. Revenue reality check (the user's actual goal)

The user's target: $2–4k/month, which was reframed as one-time debt payoff, then as ongoing. Current revenue: $0 across all three. The dependency chain to first revenue:

User: GAM service account (they offered it — never delivered)
User: MCM onboarding of helpme.tips + bestdadjokes.lol
Agent: ad wrapper integration (slots exist; spec pending)
User: traffic-buying account + float for funded tests
Agent: run tests, measure, scale or kill

Until 1–2 land, the only revenue paths are social distribution (blocked on social accounts) and organic SEO (months). Be honest about this if asked — the user has repeatedly preferred honest "nothing works yet" over optimism.

5.7 KiB Raw Blame History Unescape Escape