Websites & Cloudflare Pages¶

Most websites are deployed via Cloudflare Pages (npx wrangler pages deploy <dir> --project-name=<name>); some domains are served by standalone Cloudflare Workers instead — see Standalone Cloudflare Workers below.

Rig docs single source of truth: topology, auto-deploy, linking, and rollout for all rig documentation is maintained at https://docs.rig.dashecorp.com/docs-architecture/. Read that before editing rig doc plumbing — the section below is just the Cloudflare-level summary.

Cloudflare Pages Projects¶

Project	Custom Domain	Purpose	Source Repo	Last Modified
dashecorp	dashecorp.com, www.dashecorp.com	Dashecorp main site — app pages, blog, per-app landing pages	dashecorp/dashe-website (`website/`)	Active
invotek-no	invotek.no, www.invotek.no	Invotek AS company website	dashecorp/invotek-no (rig-managed; a merge deploys to `staging.invotek.no` behind Access, and production waits for the operator's promotion, see the repo's `docs/deploy.md`). `Stig-Johnny/invotek-no-2023` is a stale 2023 version	Active
stigjohnny-no	www.stigjohnny.no (the apex 302s to `www`; the zone is in a separate personal Cloudflare account)	Stig-Johnny's consulting profile site: CV, The Rig, FAQ	dashecorp/stigjohnny-no (rig-managed since 2026-10-02; `deploy.yml` deploys on merge; project under OpenTofu with `prevent_destroy`, dashecorp/infra#544)	Active
nutri-e	none left (its domain was to lapse 2026-10-05, claude-3#975; Domeneshop auto-renewed it on 2026-10-06 until 2027-10-05, and it is kept until `post@nutrieapp.com` is deleted (operator-inbox#42); Nutri-E 1.0.4 and older still link to it, so its paths get 301s to dashecorp.com (dashecorp/infra#646))	Former Nutri-E app website; the pages now live at www.dashecorp.com/nutri-e/ in the dashecorp project	—	Being removed
cuti-e	cuti-e.com	Cuti-E platform marketing site	—	Active
cutie-admin-dashboard	app.cuti-e.com	Cuti-E admin dashboard (customer conversations, analytics)	—	Active
cutie-super-admin	cutie-super-admin.pages.dev	Cuti-E super admin panel	—	Active
scout	scout.dashecorp.com	Scout tool	—	Active
invotek-docs	invotek-docs.pages.dev	Internal documentation site	—	Active
infra-registry	infra-registry.pages.dev	This infrastructure registry	Manual deploy from `/tmp/infra-deploy/`	Active
rig-docs	docs.rig.dashecorp.com (backend: rig-docs.pages.dev)	Rig MkDocs aggregator — pulls each rig repo's `docs/` hourly + on dispatch. Hosts `/brain/`, `/docs-architecture/`, `/components/<repo>/`.	dashecorp/rig-gitops (`docs-site/`, build-docs.sh)	Active
rig-research	research.rig.dashecorp.com (backend: rig-research.pages.dev)	Astro Starlight — BRAIN.md, `/brain/`, `/map/`, research, proposals, user-stories, whitepapers	dashecorp/rig-docs	Active
dashecorp-docs	dashecorp-docs.pages.dev	Dashecorp internal docs	dashecorp/dashe-docs	Active
run-the-docs	run-the-docs.pages.dev	Run the Docs YouTube production	run-the-docs/website	Active
ai-accountant	ai-accountant-yl9.pages.dev	AI accountant experiment	—	Active
atl-docs	atl-docs.pages.dev	ATL-E docs (legacy — agent retired)	Stig-Johnny/atl-agent (`docs/`)	Stale
gran-canaria-plan	gran-canaria-plan.pages.dev	Travel planning (personal)	—	Active

Unified rig domain (post-consolidation 2026-04-21)¶

Canonical URLs — always link these, never the *.pages.dev backends:

URL	Role
`rig.dashecorp.com/`	Landing page (Worker-rendered, managed in `dashecorp/infra` → `cloudflare/dashecorp.com/workers/rig-router.js`)
`rig.dashecorp.com/docs/*`	301 → `docs.rig.dashecorp.com/*` (legacy compat)
`docs.rig.dashecorp.com/`	MkDocs aggregator — engineering reference. Hosts `/brain/`, `/docs-architecture/`, `/components/<repo>/`, `/whitepaper/`, `/architecture/`, `/guides/`
`research.rig.dashecorp.com/`	Astro Starlight. Hosts `/brain/`, `/BRAIN.md`, `/map/`, `/research/`, `/proposals/`, `/user-stories/`, `/whitepapers/`

Retired 2026-04-21: conductor-e-docs.pages.dev, mac-executor-docs.pages.dev — their content now lives under docs.rig.dashecorp.com/components/rig-conductor/ and docs.rig.dashecorp.com/components/rig-agent-runtime/.

Plumbing, topology, and rollout log: https://docs.rig.dashecorp.com/docs-architecture/.

Dashecorp Website Structure¶

The dashecorp.com site serves as the hub for all apps. Structure in dashecorp/dashe-website/website/:

website/
├── index.html          — Main landing page
├── codi-e/             — Codi-E blog: index + 11 post directories
├── count-e/            — Count-E app page (+ privacy, support, terms)
├── drink-e/            — Drink-E app page (+ privacy, roadmap, support, terms)
├── fast-e/             — Fast-E app page (+ privacy, support, terms)
├── heart-e/            — Heart-E app page (+ privacy, roadmap, support, terms)
├── nutri-e/            — Nutri-E app page (+ privacy, roadmap, support, terms)
├── reward-e/           — Reward-E: privacy, support, terms ONLY — no product page
├── rig/                — /rig/ section: landing, agents, pipeline
├── rig-sandbox/        — unlinked + noindex page the rig flips end to end (dashe-website#209)
├── privacy/            — corporate privacy policy
├── llms.txt            — AI-readable site index
├── sitemap.xml         — SEO sitemap
├── feed.xml            — RSS feed
├── robots.txt, 404.html, styles.css, script.js, favicons
└── images/             — Shared assets

`llms.txt` and `sitemap.xml` are hand-maintained, and both are stale¶

Verified 2026-09-23: both files were last changed in the same commit (2026-09-05, PR #195) and both omit the same 18 live, indexable pages — all of Heart-E (unlisted since 2026-05-16) and Drink-E, 6 of the 11 blog posts (how-the-rig-reviews-code, watchdog-pattern, soul-and-heartbeat, github-apps-for-agents, discord-agent-communication, switching-to-claude-subscription), the /codi-e/ blog index, and /nutri-e/roadmap/. So the site advertises 5 apps while serving 7, and the sitemap costs real search visibility.

Two rules follow:

Treat the page tree as the truth, not either index — they drift in lockstep, so one cannot check the other. Walk gh api repos/dashecorp/dashe-website/contents/website/<dir>.
Adding a page means editing both files by hand today. Generating them at build time is the fix; until then, expect drift.

rig-sandbox/ is correctly absent from both — it is deliberately unlinked and noindex.

Rig-managed site repos: how a change ships¶

Applies to dashecorp/stigjohnny-no and dashecorp/invotek-no (verified 2026-10-02; merging and deploying re-verified 2026-10-23).

File an issue in the site repo and Planner-E and Dev-E pick it up; Review-E reviews the PR. Track orchestrator-only work (for example importing files only a workspace has) in an issue in dashecorp/infra or Stig-Johnny/claude-3 instead, and close it cross-repo with Closes owner/repo#N; an issue in the site repo makes Dev-E open a competing PR.
MergeGate merges a PR seconds after Review-E approves it. Both repos have had a pr-check.yml since 2026-10-14 (stigjohnny-no#24, invotek-no#33), because MergeGate merges nothing whose head has no check run. stigjohnny-no#32 and #34 and invotek-no#42 and #44 merged 9 to 13 seconds after the approval.
What a merge deploys:
stigjohnny-no: production. A PR Review-E approves goes live with no operator step.
invotek-no: staging.invotek.no, behind Access, smoke-tested. Production waits for the operator to approve the run's «Promote to production» job in GitHub (invotek-no#46, 2026-10-22). Moving that promotion into prosjekt.invotek.no, customers included, is planned (dashecorp/invotek#79).
- One production queue. The concurrency group invotek-no-production doesn't cancel runs in progress, so an older run waiting for approval holds every newer one. Its status is pending, and pending_deployments is empty.
- On 2026-10-23, PR #46's run had waited since the day before, and PR #50's run queued behind it. The operator cancelled the old run and approved the new one.
- Tell the operator the run numbers, oldest first.
- No phone notification. Runs show the operator's login as their actor, so GitHub sends no notification. Put the run links in an operator-inbox item (operator-inbox#58).
dev-e-bot cannot create or change .github/workflows/* (no GitHub App workflows permission). A human or orchestrator commits workflow files to the PR branch.
Review-E's size gate flags large imports; a genuine import gets the large-pr-ok label plus a one-line justification in the PR body.
Copy: no dashes or hyphens in visible text (names excepted). stigjohnny-no's styles.css is a purged Tailwind build, so reuse existing classes.

Consulting sites: booking, search, accessibility, analytics¶

State of www.stigjohnny.no and invotek.no as of 2026-10-13:

Booking: /book on both sites is a 302 in _redirects to Microsoft Bookings with me for post@stigjohnny.no. The public meeting type "Introsamtale / Intro call" is 30 min on Teams, Monday to Friday 10:00 to 15:00 Oslo time, with 24 h notice, 30 days ahead and a 15 min buffer. The booking link is also in that mailbox's email signature. Hours and rules are changed in Outlook, Bookings with me.
Google Search Console: URL prefix properties https://www.stigjohnny.no/ and https://invotek.no/ under invotekas@gmail.com (account /u/1/ in that browser; /u/0/ is a different account). Each is verified by an HTML meta tag in the site's index page (stigjohnny-no#14, invotek-no#25), and each sitemap is submitted. The Bing Webmaster import from Search Console is an operator step and is not done yet.
Accessibility: WCAG 2.0 AA audits on 2026-10-13 found 6 failures on stigjohnny.no and 5 on invotek.no. All are fixed (stigjohnny-no#17, invotek-no#27), and the audit notes are in each repo's docs/.
Analytics: Cloudflare Web Analytics (cookieless) was chosen over Google Analytics, so the sites keep their no cookies, no tracking line (dashecorp/infra docs/2026-10-13-web-analytics.md). The two sites are declared in cloudflare/dashecorp.com/web-analytics.tf but kept off with web_analytics_enabled = false (infra#630), because creating them needs Account Settings: Edit on the infra CI token, which has Read. Changing the token is an operator step. Then a one-line PR turns the flag on, its apply prints web_analytics_site_tokens, and each site adds the beacon script (infra#625). Until then neither site loads analytics.

Domain → Pages Mapping¶

Domain	CF Pages Project	CF Workers on same domain
dashecorp.com	dashecorp	—
invotek.no	invotek-no	invotek-account-deletion, invotek-privacy-policy, grafana tunnel
stigjohnny.no	stigjohnny-no (`www` only; apex redirect lives in the personal account's zone)	—
nutrie.app	— (Workers only)	nutrie-dsld, nutrie-openai, nutrie-apple-webhook, nutrie-stats-rss
cutiefeedback.com	— (Workers only)	cutie-worker, cutie-worker-sandbox
cuti-e.com	cuti-e (marketing), cutie-admin-dashboard (app subdomain)	—
scout.dashecorp.com	scout	—

Private/Personal Sites¶

Site	Repo	Description
Stig-Johnny CV	Stig-Johnny/cv	CV/resume (English + Norwegian)
private-sites	Stig-Johnny/private-sites	Contains: moving-planner

Standalone Cloudflare Workers (non-Pages)¶

Worker	Source Repo	Purpose
email-inbox	dashecorp/dashe-email-worker	Cloudflare Email Routing → D1 ingest for agent inboxes (`agents+*@dashecorp.com`). Deploy on merge via wrangler.
rig-router	dashecorp/infra (Tofu-managed)	`rig.dashecorp.com/` landing page + 301 of `/docs/*` to `docs.rig.dashecorp.com`. Subdomain traffic (`docs.rig` / `research.rig`) bypasses the Worker and goes straight to Pages.

Both are managed through git/CI — do not deploy manually from a laptop.

Naming convention¶

rig-* — engineering rig repos (agents, orchestration, memory)
dashe-* — shared dashecorp platform (new: dashe-email-worker)
Product names (nutri-e, cuti-e, star-rewards, ...) — one per app
Unprefixed — only for genuinely cross-cutting repos. infra is the canonical example (IaC for both rig and dashe).

Rollout tracked in dashecorp/infra#99.

Deployment¶

All Cloudflare Pages projects use manual deploy (no git integration):

# Deploy a project
npx wrangler pages deploy <directory> --project-name=<name> --branch=main

# Example: deploy dashecorp website
npx wrangler pages deploy repos/dashecorp/website --project-name=dashecorp --branch=main

# Example: deploy infra registry
npx wrangler pages deploy /tmp/infra-deploy --project-name=infra-registry --branch=main