fix(sites): SSRF hardening — validate upstream + externalUrl reject private/reserved hosts (DC-074) [glm-grade=A]

Pre-fix, an authenticated dashboard operator could call:
  POST /api/v1/site         {domain:"evil.example.com", upstream:"10.0.0.1:80"}
  POST /api/v1/site/external {subdomain:"x", externalUrl:"http://192.168.1.5"}
and end up with a Caddy site block that proxies PUBLIC traffic at
evil.example.com to an INTERNAL host. Caddy runs on DNS2 (same
network as the targets), so the SSRF lands.

The pre-fix /site upstream regex /^[a-z0-9.-]+:\d{1,5}$/i only
checked charset — it happily accepted 192.168.1.1:80 and
169.254.169.254:80 (AWS metadata IP). /site/external called
validateURL() without blockPrivate:true, leaving the door wide open.

(1) New helper validateUpstream() in fleet-validation.js — reuses
    resolveAndCheckAddress() (DC-068 SSRF work) to reject literal
    private IPv4/IPv6 (loopback / RFC1918 / link-local / CGNAT /
    multicast / broadcast / 0.0.0.0 / TEST-NET / benchmark ranges),
    resolve hostnames and reject private answers (rebinding defense),
    and cap port to 1..65535. Opt-in via SITES_ALLOW_PRIVATE_UPSTREAMS=true.

(2) /site calls validateUpstream() BEFORE caddy.modify() — gate
    happens before any state mutation. Throws ValidationError with
    canonical [DC-074] tag and a redacted hostname audit log entry.

(3) /site/external calls validateURL() (syntax only) + validateUpstream()
    (private-IP gate). validateURL's blockPrivate is intentionally
    NOT passed because it has no opt-in — that's what validateUpstream
    is for.

(4) Tests (__tests__/routes/sites-dc074.routes.test.js, NEW, 60/60
    passing): helper unit tests (format, literal IPv4/IPv6 private
    reject, public IP accept, hostname resolve + rebinding defense,
    env opt-in override), POST /site integration (10 regression
    payloads + public accept + opt-in + port range + charset), POST
    /site/external integration (8 regression payloads + public
    accept + DNS rebinding defense + opt-in), canonical SSRF regression
    proof (RFC 1918 literal IPv4 in upstream + RFC 1918 literal IPv4
    in URL host), unchanged-behavior checks on isPrivateOrReservedIPv4/IPv6.

Full repo suite: 2402/2402 tests in 102 suites (zero regressions).
GLM-5.3 stand-in judge round 1 (deleg_384b9f53, 41.46s, 3 tool
calls, MiniMax-M3 per Sami authorization 2026-08-17): A ship-first.

Refs: codex-as-judge SKILL.md 'Stand-in fallback chain'. Verdict
record: /root/dashcaddy-polish/.ump-verdicts/2026-08-18T22-35-00Z-dc-074-round-1-A.json

dashcaddy-api/src/utilities/fleet-validation.js

@@ -415,10 +415,91 @@ function validateFleetHost(input) {
   };
 }
 
+/**
+ * Validate a `host:port` upstream string for use in Caddy's `reverse_proxy`.
+ *
+ * DC-074 SSRF hardening: an authenticated dashboard operator can call
+ * POST /api/v1/site with `upstream: '10.0.0.1:80'` and end up with a
+ * Caddyfile entry that proxies public traffic (https://attacker.example.com)
+ * to an INTERNAL host (10.0.0.1:80). Caddy runs on DNS2 — same network
+ * as the targets — so the proxy lands the request on the private host.
+ * The operator doesn't even need DNS-rebinding tricks: a literal IPv4
+ * like 192.168.1.1 is accepted by the existing `[a-z0-9.-]+:\d{1,5}`
+ * upstream regex.
+ *
+ * Reuses `resolveAndCheckAddress()` to:
+ *   - reject literal private IPv4 / IPv6
+ *   - resolve DNS names and reject any private-IP answer
+ *     (rebinding defense — the actual address Caddy connects to is
+ *     the resolved IP at registration time; Caddy itself resolves
+ *     the name per-request, so a malicious operator could flip the
+ *     A record between registration and connection. Acceptable
+ *     residual risk — the registration check is the main gate.)
+ *   - cap port to 1..65535 (defense vs. `host:99999999` integer
+ *     overflow / Caddy parser-bomb)
+ *
+ * Opt-in via SITES_ALLOW_PRIVATE_UPSTREAMS=true for operators who
+ * intentionally proxy to private targets (faster than a public DNS
+ * round-trip + central control plane).
+ *
+ * @param {string} upstream - "host:port" string (e.g. "10.0.0.1:80")
+ * @param {object} [opts]
+ * @param {boolean} [opts.allowPrivate] - override the env-var default
+ * @returns {Promise<{ok: true, host: string, port: number, resolvedIp?: string, family?: number} | {ok: false, code: string, message: string}>}
+ */
+async function validateUpstream(upstream, opts = {}) {
+  if (typeof upstream !== 'string' || upstream.length === 0) {
+    return { ok: false, code: 'INVALID_UPSTREAM', message: 'upstream is required' };
+  }
+
+  // Split on the LAST colon so IPv6 literals like `[::1]:80` parse
+  // correctly (and a malformed `[::1]` without port is rejected with
+  // a clean code, not a confusing TypeError from Number()).
+  const lastColon = upstream.lastIndexOf(':');
+  if (lastColon < 0) {
+    return { ok: false, code: 'INVALID_UPSTREAM', message: 'upstream must be host:port' };
+  }
+  const host = upstream.slice(0, lastColon);
+  const portStr = upstream.slice(lastColon + 1);
+
+  const portNum = Number(portStr);
+  if (!Number.isInteger(portNum) || portNum < 1 || portNum > 65535) {
+    return { ok: false, code: 'INVALID_PORT', message: 'upstream port must be an integer 1..65535' };
+  }
+
+  // Allow-list the host charset BEFORE the DNS lookup so attacker
+  // payloads can't make the resolver do work. Matches the fleet
+  // isValidHostnameSyntax check; sites.js's own `[a-z0-9.-]+` regex
+  // is more restrictive (only letters/digits/dots/hyphens) so
+  // we widen here to also accept bracketed IPv6. Anything else gets
+  // rejected pre-DNS.
+  const isBracketedIPv6 = host.startsWith('[') && host.endsWith(']');
+  const hostToCheck = isBracketedIPv6 ? host.slice(1, -1) : host;
+  if (!isValidHostnameSyntax(hostToCheck) && require('net').isIP(hostToCheck) === 0) {
+    return { ok: false, code: 'INVALID_HOST', message: `upstream host "${host}" is not a valid DNS name or IP address` };
+  }
+
+  const allowPrivate = typeof opts.allowPrivate === 'boolean'
+    ? opts.allowPrivate
+    : process.env.SITES_ALLOW_PRIVATE_UPSTREAMS === 'true';
+
+  const r = await resolveAndCheckAddress(hostToCheck, { allowPrivate });
+  if (!r.ok) return r; // bubbles up PRIVATE_IPV4 / PRIVATE_IPV6 / INVALID_HOSTNAME / DNS_*
+
+  return {
+    ok: true,
+    host,
+    port: portNum,
+    resolvedIp: r.ip,
+    family: r.family,
+  };
+}
+
 module.exports = {
   validateFleetHost,
   resolveAndCheckAddress,
   isPrivateOrReservedIPv4,
   isPrivateOrReservedIPv6,
   isValidHostnameSyntax,
+  validateUpstream,
 };