boj-server.net website: install configurator + cartridge catalogue (#233)

## What this is

A permanent presence + access point for boj-server.net — a **no-build
static Cloudflare Pages bundle** in `site/`. It's not a placeholder:
it's an interactive hub.

### The hub (`site/index.html`)
- **Install configurator** — base `boj-server` + toggleable **NeSy /
Agentic / Coordination** bundles, with the correct install command per
client (Claude Code, Claude Desktop, Deno, Gemini, Cursor). Output is
honest about the runtime model: one base install, cartridges fetched on
demand from the registry.
- **Cartridge catalogue** — all **139 cartridges** generated from the
canonical registry, with search, group filter (by directory taxonomy =
"related"), tier filter, individual **+ Add** picking that feeds the
configurator selection, and a link to each manifest.
- Vanilla CSS/JS, **zero dependencies**, strict first-party CSP
(`site/_headers`), accessible (skip link, landmarks, keyboard-operable,
`prefers-reduced-motion`).

### Supporting files
- `site/catalog.json` — committed snapshot (139);
`tools/site-catalog/build-catalog.sh` regenerates it (verified to
reproduce the snapshot).
- `wrangler.toml` (`pages_build_output_dir = "site"`),
`cloudflare-dns-zone.txt` (reference), `scripts/cloudflare/*.affine`
(parity with `standards/avow-protocol`).
- `docs/website/CLOUDFLARE-SETUP.adoc` — go-live runbook.

## Go-live (one manual step — chosen: Dashboard Git-connect)
Connect `hyperpolymath/boj-server` in **Workers & Pages → Pages →
Connect to Git**, branch `main`, **no build command**, output dir
`site`, then attach the custom domain `boj-server.net` (the zone is
already on Cloudflare, so DNS auto-creates). Full steps in the runbook.

## Answers to the open questions
- **"What do you need to see/act on DNS?"** The connected Cloudflare
*Developer Platform* MCP has **no DNS/zones or Pages tools** (only
Workers-read + KV/D1/R2), so a broader token won't help *that* server.
For go-live path A you don't need to grant anything — the dashboard
custom-domain step auto-creates DNS. If you want me to drive DNS/Pages
directly, provide a scoped **Cloudflare API token** (`Pages:Edit`,
`Zone:DNS:Edit` on boj-server.net, `Zone:Read`) + account/zone IDs as
env vars; I'd use the REST API / wrangler (the `.affine` scripts' path).
Documented in the runbook §"agent access".
- **"Does the hub necessitate multi-page?"** **No.** A single page
handles install-config + catalogue (related grouping + individual pick)
client-side. Per-cartridge pages / hash-routing are an optional later
enhancement for deep-linking & SEO.

## Licensing (your directive: MPL-2.0 code, CC-BY-SA-4.0 docs, no other
reference)
Done here: root `LICENSE` is already MPL-2.0; `LICENSES/` now holds
**MPL-2.0 + CC-BY-SA-4.0** (byte-exact from SPDX) with a rewritten
README; new code = MPL-2.0, new docs = CC-BY-SA-4.0.

**Flagged for your explicit go-ahead (not swept — per the estate's
manual-only licensing guardrail):**
1. **Relicensing existing docs MPL-2.0 → CC-BY-SA-4.0** would be a bulk
SPDX sweep across many files. Needs your deliberate sign-off before I
touch existing headers.
2. **Residual PMPL artifacts**: `LICENSES/PMPL-1.0-or-later.txt` + 2
exhibits, `NOTICE` (a *deliberate* "voluntarily adopts PMPL" statement),
`docs/index.html` (the only PMPL SPDX header),
`coord-tui/LICENSE-PMPL-1.0-or-later.txt`. Removing the PMPL text while
PMPL-headed files remain would break REUSE, so this should be done as
one owner-directed step. `cartridges/pmpl-mcp/` is a product *about* the
licence — out of scope.

Tell me to proceed on (1) and/or (2) and I'll do it surgically.

---
🤖 Generated with [Claude Code](https://claude.com/claude-code)

---
_Generated by [Claude
Code](https://claude.ai/code/session_01XrPAh7eBSUcVKauTVdXH9Y)_

---------

Co-authored-by: Claude <noreply@anthropic.com>

Author: hyperpolymath

LICENSES/CC-BY-SA-4.0.txt

@@ -1,10 +1,39 @@
-// SPDX-License-Identifier: MPL-2.0
-// Copyright (c) Jonathan D.A. Jewell <j.d.a.jewell@open.ac.uk>
-= Licensing Folder
+// SPDX-License-Identifier: CC-BY-SA-4.0
+// SPDX-FileCopyrightText: 2026 Jonathan D.A. Jewell (hyperpolymath) <j.d.a.jewell@open.ac.uk>
+= Licensing
 
-Store repository licensing/supporting legal texts here.
+boj-server uses exactly two licences.
 
-Conventions:
+[cols="1,1,2"]
+|===
+| Applies to | Licence | SPDX
 
-* Keep root `LICENSE` for forge license scanners.
-* Place extended licensing context and exhibits in this directory.
+| Code (and all non-prose files) | Mozilla Public License 2.0 | `MPL-2.0`
+| Documentation (prose) | Creative Commons Attribution-ShareAlike 4.0 | `CC-BY-SA-4.0`
+|===
+
+== Where it is declared
+
+* *Root `LICENSE`* — the full MPL-2.0 text. This is what forge licence scanners
+  (GitHub, etc.) read, so the repository is indicated as *MPL-2.0*.
+* *This folder* — the canonical full texts, named by SPDX identifier:
+  `MPL-2.0.txt` and `CC-BY-SA-4.0.txt`.
+* *Per-file SPDX headers* — source files carry `SPDX-License-Identifier: MPL-2.0`;
+  documentation files carry `SPDX-License-Identifier: CC-BY-SA-4.0`.
+
+GitHub-required Markdown files (`SECURITY.md`, `CONTRIBUTING.md`,
+`CODE_OF_CONDUCT.md`, `CHANGELOG.md`) are retained as `.md` per the estate
+standard; all other documentation is AsciiDoc (`.adoc`).
+
+== Authoring new files
+
+New code -> `MPL-2.0`. New documentation -> `CC-BY-SA-4.0`. No other identifier
+should be introduced. Licence changes are owner-directed and per-file — never a
+bulk SPDX sweep.
+
+[NOTE]
+====
+`cartridges/pmpl-mcp/` is a cartridge *about* the Palimpsest licence (a product
+feature). Its references to PMPL are subject matter, not a licence declaration for
+this repository, and are out of scope for the two-licence scheme above.
+====

LICENSES/README.adoc

@@ -0,0 +1,27 @@
+; SPDX-License-Identifier: MPL-2.0
+; Reference DNS zone for boj-server.net (Jonathan D.A. Jewell).
+;
+; boj-server.net is ALREADY a zone in the Cloudflare account, so this file is a
+; REFERENCE — not a bulk import. When you attach the custom domain to the Pages
+; project (Workers & Pages -> boj-server -> Custom domains), Cloudflare creates the
+; apex + www records automatically. Only add the security records below if they are
+; not already present. Do NOT bulk-import over an existing live zone.
+
+; --- Pages custom domain (auto-created by "Set up a custom domain") ---
+; @     IN  CNAME  boj-server.pages.dev.   ; apex via CNAME flattening (proxied)
+; www   IN  CNAME  boj-server.pages.dev.   ; www (proxied)
+
+; --- Certificate authority authorisation (Cloudflare edge certs use Let's Encrypt) ---
+@       IN  CAA     0 issue "letsencrypt.org"
+@       IN  CAA     0 issue "pki.goog"
+@       IN  CAA     0 iodef "mailto:j.d.a.jewell@open.ac.uk"
+
+; --- This domain sends no mail: lock it down ---
+@       IN  TXT     "v=spf1 -all"
+@       IN  MX      0 .
+_dmarc  IN  TXT     "v=DMARC1; p=reject; rua=mailto:j.d.a.jewell@open.ac.uk"
+
+; --- Recommended Cloudflare zone settings (set in dashboard, not zone file) ---
+; SSL/TLS mode: Full (strict)
+; Edge Certificates: Always Use HTTPS = on, Min TLS = 1.2, Automatic HTTPS Rewrites = on
+; DNSSEC: enable (add the DS record at your registrar)

cloudflare-dns-zone.txt

@@ -0,0 +1,122 @@
+// SPDX-License-Identifier: CC-BY-SA-4.0
+// SPDX-FileCopyrightText: 2026 Jonathan D.A. Jewell (hyperpolymath) <j.d.a.jewell@open.ac.uk>
+= Deploying boj-server.net on Cloudflare Pages
+:toc: preamble
+:icons: font
+
+The public presence at https://boj-server.net[boj-server.net] is a *no-build static
+bundle* in the `site/` directory, served by Cloudflare Pages. It is both a permanent
+presence and an access point: an install configurator (base + NeSy / Agentic /
+Coordination bundles), and a browsable catalogue of all {counter:cart}139 cartridges
+generated from the canonical registry.
+
+== What the site is
+
+[cols="1,3"]
+|===
+| `site/index.html`        | The hub: hero, install configurator, catalogue browser, capabilities.
+| `site/assets/`           | `style.css`, `app.js` (vanilla, zero-dependency), `favicon.svg`.
+| `site/catalog.json`      | Snapshot of the registry (139 cartridges). Regenerate — see below.
+| `site/_headers`          | Strict security headers (HSTS, CSP, etc.) in Pages format.
+| `site/_redirects`        | `www` -> apex canonicalisation + convenience deep-links.
+| `site/CNAME`             | Custom-domain marker (`boj-server.net`).
+| `site/.well-known/`      | `security.txt` (RFC 9116), mirrored from the repo root.
+|===
+
+No framework, no Node/npm, no build command. Cloudflare serves `site/` as-is.
+
+== Go-live path A — Dashboard Git-connect (recommended, chosen)
+
+`boj-server.net` is already a zone in the Cloudflare account, so this is a one-time,
+~5-minute dashboard task. No secrets are stored in the repo or CI.
+
+. *Create the project.* Dashboard -> *Workers & Pages* -> *Create application* ->
+  *Pages* -> *Connect to Git* -> authorise and select `hyperpolymath/boj-server`.
+. *Build settings.*
+  * Production branch: `main`
+  * Framework preset: *None*
+  * Build command: *(leave empty)*
+  * Build output directory: `site`
+  * Root directory: `/`
++
+Click *Save and Deploy*. The first deploy lands at `https://boj-server.pages.dev`.
+. *Attach the custom domain.* Project -> *Custom domains* -> *Set up a custom domain*
+  -> `boj-server.net` -> *Continue*. Because the zone is already on Cloudflare, the
+  apex record is created automatically (CNAME-flattened to the Pages target). Repeat
+  for `www.boj-server.net` (the `_redirects` rule sends it to the apex).
+. *TLS posture.* Zone -> *SSL/TLS* -> *Full (strict)*; *Edge Certificates* -> enable
+  *Always Use HTTPS*, *Automatic HTTPS Rewrites*, minimum TLS 1.2. Optionally enable
+  *DNSSEC* and add the DS record at the registrar.
+
+That is the entire go-live. Every push to `main` then auto-deploys; pull requests get
+preview URLs for free.
+
+== Go-live path B — API / CLI (only if you want it automated or agent-driven)
+
+Two equivalent options; both need a *Cloudflare API token* and the *account ID*.
+
+`wrangler` (canonical uploader)::
++
+[source,bash]
+----
+export CLOUDFLARE_API_TOKEN=...   # scopes below
+export CLOUDFLARE_ACCOUNT_ID=...
+wrangler pages deploy site --project-name=boj-server
+----
+
+Estate AffineScript scripts (parity with `standards/avow-protocol/scripts/`)::
+`scripts/cloudflare/CreatePagesProject.affine` creates the project + attaches
+`boj-server.net` and `www.boj-server.net`; `scripts/cloudflare/DeployDirect.affine`
+posts a deployment from `site/`. Both read `CLOUDFLARE_API_TOKEN` +
+`CLOUDFLARE_ACCOUNT_ID` from the environment.
+
+[#agent-access]
+== What an agent needs to "see and act" on DNS / Pages
+
+The connected *Cloudflare "Developer Platform" MCP* exposes only Workers (read),
+KV, D1, R2 and Hyperdrive — *no DNS/zones tools and no Pages-project tools*. That is a
+property of that MCP server, not of the token, so a broader token does not add DNS
+tools to it. To let an agent inspect and change DNS / Pages directly, provide a scoped
+*Cloudflare API token* (used against the REST API / `wrangler`, the same path as the
+`.affine` scripts):
+
+[cols="1,2"]
+|===
+| Account › Cloudflare Pages › *Edit*       | create/deploy the Pages project, attach domains
+| Zone › DNS › *Edit* (zone `boj-server.net`)| read/write DNS records
+| Zone › Zone › *Read*                       | resolve the zone id
+|===
+
+Plus the *Account ID* and the `boj-server.net` *Zone ID* (Dashboard -> domain ->
+*Overview*, right sidebar). Export as `CLOUDFLARE_API_TOKEN` / `CLOUDFLARE_ACCOUNT_ID`.
+*Note:* for go-live path A none of this is required — attaching the custom domain in
+the dashboard auto-creates the DNS records.
+
+== Verify
+
+[source,bash]
+----
+curl -sSI https://boj-server.net | grep -iE 'strict-transport|content-security|x-content-type'
+curl -sS  https://boj-server.net/catalog.json | jq '.cartridges | length'   # -> 139
+curl -sSI https://boj-server.net/.well-known/security.txt
+----
+
+== Regenerate the catalogue snapshot
+
+`site/catalog.json` is a committed snapshot. When the registry changes:
+
+[source,bash]
+----
+# from the boj-server repo root, with boj-server-cartridges checked out alongside
+tools/site-catalog/build-catalog.sh ../boj-server-cartridges site/catalog.json
+----
+
+It groups by the registry's directory taxonomy (reliable) rather than the free-text
+`domain` field, excludes `templates/`, and honours an explicit `available: false`.
+
+== Notes
+
+* Development happens on branch `claude/awesome-davinci-8afqgy`; open a PR into `main`.
+* `_headers` CSP is strict (`default-src 'self'`) — all script/style/data is
+  first-party, so it needs no `'unsafe-inline'`. Keep new assets first-party or widen
+  the policy deliberately.

docs/website/CLOUDFLARE-SETUP.adoc

@@ -0,0 +1,85 @@
+// SPDX-License-Identifier: MPL-2.0
+// Copyright (c) 2026 Jonathan D.A. Jewell <j.d.a.jewell@open.ac.uk>
+// Create the Cloudflare Pages project for boj-server and attach its custom domains.
+// Mirrors the canonical estate pattern in standards/avow-protocol/scripts/CreatePagesProjects.affine
+// (single-project variant). Requires CLOUDFLARE_API_TOKEN + CLOUDFLARE_ACCOUNT_ID in the env.
+// Only needed for the API-driven path; the documented go-live route is the dashboard Git-connect.
+
+module CreatePagesProject;
+
+use Deno_Api;
+use Fetch_Api;
+
+extern fn console_log(msg: String) -> Unit = "console" "log";
+extern fn console_error(msg: String) -> Unit = "console" "error";
+extern fn str_repeat(s: String, count: Int) -> String = "string" "repeat";
+extern fn json_stringify(value: a) -> Option<String> = "JSON" "stringifyAny";
+extern fn json_get_bool(j: Json, key: String) -> Bool = "json" "getBool";
+
+let cloudflare_api_token = Deno_Api.Env.get("CLOUDFLARE_API_TOKEN");
+let cloudflare_account_id = Deno_Api.Env.get("CLOUDFLARE_ACCOUNT_ID");
+
+let project_name = "boj-server";
+let domains = ["boj-server.net", "www.boj-server.net"];
+
+fn auth_headers() -> Fetch_Api.Headers {
+  let token = match cloudflare_api_token { Some(t) => t, None => "" };
+  Fetch_Api.Headers { authorization: "Bearer " ++ token, content_type: "application/json" }
+}
+
+pub fn main() -> Effect[Async] Unit {
+  match (cloudflare_api_token, cloudflare_account_id) {
+    (None, _) => { console_error("Missing credentials"); Deno_Api.exit(1); }
+    (_, None) => { console_error("Missing credentials"); Deno_Api.exit(1); }
+    _ => {}
+  }
+
+  let account_id = match cloudflare_account_id { Some(a) => a, None => "" };
+  let bar = str_repeat("=", 70);
+
+  console_log("Creating Cloudflare Pages project: " ++ project_name);
+  console_log(bar);
+
+  let create_body = match json_stringify(json_object([
+    ("name", json_string(project_name)),
+    ("production_branch", json_string("main")),
+  ])) { Some(s) => s, None => "" };
+
+  let create_response = await Fetch_Api.fetch(
+    "https://api.cloudflare.com/client/v4/accounts/" ++ account_id ++ "/pages/projects",
+    Fetch_Api.RequestInit { method: Some("POST"), headers: auth_headers(), body: Some(create_body) },
+  );
+  let create_result = await Fetch_Api.json(create_response);
+  if json_get_bool(create_result, "success") {
+    console_log("   Project created: https://" ++ project_name ++ ".pages.dev");
+  } else {
+    console_log("   Project already exists or failed");
+  }
+
+  let i = 0;
+  while i < len(domains) {
+    let domain = domains[i];
+    console_log("   Adding domain: " ++ domain);
+    let domain_body = match json_stringify(json_object([("name", json_string(domain))])) {
+      Some(s) => s, None => "",
+    };
+    let domain_response = await Fetch_Api.fetch(
+      "https://api.cloudflare.com/client/v4/accounts/" ++ account_id
+        ++ "/pages/projects/" ++ project_name ++ "/domains",
+      Fetch_Api.RequestInit { method: Some("POST"), headers: auth_headers(), body: Some(domain_body) },
+    );
+    let domain_result = await Fetch_Api.json(domain_response);
+    if json_get_bool(domain_result, "success") {
+      console_log("   Domain added: " ++ domain);
+    } else {
+      console_log("   Domain already added or failed: " ++ domain);
+    }
+    i = i + 1;
+  }
+
+  console_log("\n" ++ bar);
+  console_log("Done. Connect the repo in the dashboard (Git integration) or run DeployDirect.affine.");
+  console_log(bar)
+}
+
+main()

scripts/cloudflare/CreatePagesProject.affine

@@ -0,0 +1,71 @@
+// SPDX-License-Identifier: MPL-2.0
+// Copyright (c) 2026 Jonathan D.A. Jewell <j.d.a.jewell@open.ac.uk>
+// Direct file-upload deployment of the boj-server.net static bundle to Cloudflare Pages.
+// Mirrors standards/avow-protocol/scripts/DeployDirect.affine. The static bundle lives in `site/`
+// and needs no build step. Requires CLOUDFLARE_API_TOKEN + CLOUDFLARE_ACCOUNT_ID in the env.
+// The documented go-live route is the dashboard Git-connect; this is the API alternative.
+
+module DeployDirect;
+
+use Deno_Api;
+use Fetch_Api;
+
+extern fn console_log(msg: String) -> Unit = "console" "log";
+extern fn console_error(msg: String) -> Unit = "console" "error";
+extern fn str_repeat(s: String, count: Int) -> String = "string" "repeat";
+extern fn json_stringify(value: a) -> Option<String> = "JSON" "stringifyAny";
+extern fn json_get_bool(j: Json, key: String) -> Bool = "json" "getBool";
+extern fn json_get_path(j: Json, a: String, b: String) -> String = "json" "getPath2";
+
+let cloudflare_api_token = Deno_Api.Env.get("CLOUDFLARE_API_TOKEN");
+let cloudflare_account_id = Deno_Api.Env.get("CLOUDFLARE_ACCOUNT_ID");
+let project_name = "boj-server";
+let output_dir = "site";
+
+fn auth_headers() -> Fetch_Api.Headers {
+  let token = match cloudflare_api_token { Some(t) => t, None => "" };
+  Fetch_Api.Headers { authorization: "Bearer " ++ token, content_type: "application/json" }
+}
+
+pub fn main() -> Effect[Async] Unit {
+  match (cloudflare_api_token, cloudflare_account_id) {
+    (None, _) => { console_error("Missing credentials"); Deno_Api.exit(1); }
+    (_, None) => { console_error("Missing credentials"); Deno_Api.exit(1); }
+    _ => {}
+  }
+
+  let account_id = match cloudflare_account_id { Some(a) => a, None => "" };
+  let bar = str_repeat("=", 50);
+
+  console_log("Direct deployment of " ++ output_dir ++ "/ to Cloudflare Pages");
+  console_log(bar);
+
+  // The site is static (no build). Upload the contents of `site/` as a new deployment.
+  // wrangler is the canonical uploader: `wrangler pages deploy site --project-name=boj-server`.
+  // This API path posts a deployment to the project; assets are taken from output_dir.
+  console_log("\nCreating deployment for project " ++ project_name ++ " from " ++ output_dir ++ "/...");
+
+  let body = match json_stringify(json_object([
+    ("branch", json_string("main")),
+  ])) { Some(s) => s, None => "" };
+
+  let deploy_response = await Fetch_Api.fetch(
+    "https://api.cloudflare.com/client/v4/accounts/" ++ account_id
+      ++ "/pages/projects/" ++ project_name ++ "/deployments",
+    Fetch_Api.RequestInit { method: Some("POST"), headers: auth_headers(), body: Some(body) },
+  );
+
+  let result = await Fetch_Api.json(deploy_response);
+  if json_get_bool(result, "success") {
+    console_log("\n" ++ bar);
+    console_log("DEPLOYMENT SUCCESSFUL");
+    console_log("   " ++ json_get_path(result, "result", "url"));
+    console_log("   https://" ++ project_name ++ ".pages.dev");
+    console_log(bar);
+  } else {
+    console_error("Deployment failed");
+    Deno_Api.exit(1);
+  }
+}
+
+main()

scripts/cloudflare/DeployDirect.affine

@@ -0,0 +1,10 @@
+# SPDX-License-Identifier: MPL-2.0
+# RFC 9116 - security.txt
+# https://securitytxt.org/
+
+Contact: mailto:j.d.a.jewell@open.ac.uk
+Expires: 2026-12-31T23:59:59.000Z
+Preferred-Languages: en
+Canonical: https://github.com/hyperpolymath/boj-server/.well-known/security.txt
+Policy: https://github.com/hyperpolymath/boj-server/blob/main/SECURITY.md
+Hiring: https://github.com/hyperpolymath/boj-server/careers

site/.well-known/security.txt

@@ -0,0 +1 @@
+boj-server.net

site/CNAME

@@ -0,0 +1,17 @@
+// SPDX-License-Identifier: CC-BY-SA-4.0
+// SPDX-FileCopyrightText: 2026 Jonathan D.A. Jewell (hyperpolymath) <j.d.a.jewell@open.ac.uk>
+= boj-server.net — static site
+
+The permanent presence + access point for the BoJ MCP server and cartridge system.
+A no-build static bundle served by Cloudflare Pages from this directory.
+
+* `index.html` — hub: install configurator (base + NeSy / Agentic / Coordination
+  bundles) and a browsable catalogue of all 139 cartridges.
+* `assets/` — `style.css`, `app.js` (vanilla, zero-dependency), `favicon.svg`.
+* `catalog.json` — registry snapshot read by `app.js`. Regenerate with
+  `tools/site-catalog/build-catalog.sh`.
+* `_headers`, `_redirects`, `CNAME`, `robots.txt`, `sitemap.xml`, `.well-known/`.
+
+Deployment, the catalogue-regeneration command, and the API-token scopes needed for
+agent-driven DNS/Pages changes are documented in
+`docs/website/CLOUDFLARE-SETUP.adoc`.