diff --git a/client/src/api/client.js b/client/src/api/client.js
index 068e99b..0051b81 100644
--- a/client/src/api/client.js
+++ b/client/src/api/client.js
@@ -51,7 +51,10 @@ function safeParse(text) {
 // The `api` object below stays core's own binding surface. Its `atlas` and
 // `shard` namespaces are module bindings that only still live here because
 // Phase 3 has not moved them yet.
-export { req as request }
+// `BASE` goes with it: a module that needs an EventSource URL cannot go through
+// `req` (fetch-only) and must not hardcode `/api/v1`, which is core's choice of
+// mount point and not a promise it has made.
+export { req as request, BASE }
 
 export const api = {
   // ----- auth -----
diff --git a/client/src/main.jsx b/client/src/main.jsx
index f1c81fa..c2c3101 100644
--- a/client/src/main.jsx
+++ b/client/src/main.jsx
@@ -3,10 +3,7 @@ import { createRoot } from 'react-dom/client'
 import { BrowserRouter } from 'react-router-dom'
 import App from './App.jsx'
 import { publishSharedDependencies } from './modules/shared.js'
-import { declareSlot, registerExtension, registerFeatureProvider } from './modules/registry.js'
-import { useShardFlags } from './lib/useShardFeatures.js'
-import ShardStatusLink from './components/ShardStatusLink.jsx'
-import UserShardSections from './routes/admin/views/UserShardSections.jsx'
+import { declareSlot } from './modules/registry.js'
 import './styles/theme.css'
 
 // Publish window.__rg BEFORE rendering and before any module chunk evaluates.
@@ -17,18 +14,12 @@ import './styles/theme.css'
 // (docs/website/MODULE_API.md §3.2).
 publishSharedDependencies()
 
-// Core registers through the same seam a module uses, and registers FIRST — the
-// client twin of the server's `registries.registerCore()` (MODULE_SYSTEM.md
-// §1.9). The ten shard-gated rows in the public header are core's only because
-// Phase 3 has not moved them yet; routing them through the registry now means
-// SiteHeader holds one mechanism instead of two, and the extraction becomes a
-// deletion rather than a rewrite made under extraction pressure.
-//
-// The owner id is `core`, which is what a nav row with no `moduleId` resolves
-// against (modules/featureGate.js). The namespace is `uo`, so a module that
-// wants to read these flags — the `uo` module itself, once it owns them — asks
-// for them by the name they will always have had.
-registerFeatureProvider('core', 'uo', useShardFlags)
+// Core registered a feature provider here until slice 3, under owner id `core`
+// and namespace `uo`, so that the seam was exercised by real content from the
+// day it was built. That prediction paid out exactly as written: the extraction
+// deleted the registration and the hook it named, and SiteHeader was not touched.
+// There is nothing for core to register now — no core nav row carries a
+// `feature` — and the filter is a correct no-op until a module supplies one.
 
 // ── Extension slots (MODULE_API.md §3.7) ───────────────────────────────────
 //
@@ -48,15 +39,22 @@ declareSlot('site.footer.status')
 // resource, one extension point, two halves. The module with routes under
 // /api/v1/admin/users/:id is the module with something to show on that page.
 declareSlot('admin.users.detail')
+// The invite-acceptance page's optional next step. Core owns invites — staff are
+// invited too — and owned the game-account step inside them until slice 3, which
+// meant core reading a `gameAccountSignup` flag and posting to a shard route.
+//
+// Named for the place, like the other two: it is "the point after an invite has
+// been accepted and before the invitee is sent on", not "create a game account".
+// Whether there is a step at all is the filling module's decision, made from
+// data core does not have; core renders the shell and a skip control, and hands
+// over `onDone`. With the slot unfilled the invitee goes straight to the portal,
+// which is what core's own code did whenever the flag was off.
+declareSlot('player.invite.accepted')
 
-// And core fills both itself, under owner id `core`, with the components that
-// were inline in SiteFooter.jsx and UserDetail.jsx until this slice. The page
-// renders exactly what it rendered before, and the mechanism is exercised by
-// core's own content from the day it lands rather than first proved by the
-// change that depends on it. Slice 3 deletes these three lines and the two files
-// they name, and the module registers the same two slots on its way in.
-registerExtension('core', 'site.footer.status', ShardStatusLink)
-registerExtension('core', 'admin.users.detail', UserShardSections)
+// Core filled the first two itself until slice 3, with the components that were
+// inline in SiteFooter.jsx and UserDetail.jsx. Both are gone: the module fills
+// all three, and core's own fills had to go for it to be able to — the first
+// fill wins, and core registered first (§3.7).
 
 // Render on DOMContentLoaded rather than immediately, and that is the one line
 // of core's boot the module system changes.
diff --git a/client/src/modules/featureGate.js b/client/src/modules/featureGate.js
index dc30912..4c6091c 100644
--- a/client/src/modules/featureGate.js
+++ b/client/src/modules/featureGate.js
@@ -3,22 +3,22 @@
 // Phase 2, PR 8 of docs/website/MODULE_SYSTEM.md §2.7 (§1.5 states the problem);
 // the contract is docs/website/MODULE_API.md §3.3.
 //
-// Ten of the sixteen rows in the public header carry a `feature`, and every one
-// of them is a shard surface an admin can disable or gate to a higher audience.
-// The provider that answers those questions — `useShardFeatures` — moves out
-// with the module, so core cannot keep calling it directly and still be a core.
-// It keeps a generic seam instead, and the module fills it.
+// Nine of the sixteen rows in the public header used to carry a `feature`, and
+// every one of them was a shard surface an admin can disable or gate to a higher
+// audience. The provider that answered those questions moved out with the module
+// in Phase 3 slice 3, and core cannot call it directly and still be a core. It
+// keeps this generic seam instead, and the module fills it.
 //
 // **The namespace comes from the registration, not from the string.** A row's
 // `feature` is resolved by the provider its OWN module registered, so a module
 // author writes `feature: 'status'` exactly as it reads today: nothing parses a
 // prefix, and a typo'd namespace is not a thing that can exist. Core's own rows
-// carry no `moduleId` and resolve against the owner id `core`, which is what
-// core registers `useShardFeatures` under until Phase 3 moves those rows into
-// the module and they arrive stamped `uo` instead.
+// carry no `moduleId` and resolve against the owner id `core` — which nothing
+// registers now that the shard rows are gone, and that is the correct resting
+// state rather than a gap: no core nav row carries a `feature`.
 //
-// Everything here fails OPEN, and that is deliberate and unchanged from
-// useShardFeatures' own posture: this is presentation, the gate is server-side
+// Everything here fails OPEN, and that is deliberate: this is presentation, the
+// gate is server-side
 // (a disabled feature 404s and an out-of-rung one 403s whether or not a link was
 // rendered), so an unknown answer shows the link rather than blanking the nav.
 // The one thing a UI mistake must never do here is hide a page from someone
diff --git a/client/src/modules/registry.js b/client/src/modules/registry.js
index 87d7a03..d6783f5 100644
--- a/client/src/modules/registry.js
+++ b/client/src/modules/registry.js
@@ -79,8 +79,18 @@ export function registerRoutes(id, byArea) {
  * core group, `order` sorts within it, and an unknown group name appends rather
  * than dropping the item — a mis-typed group must cost a position, never a link.
  *
+ * `icon` is a component core renders exactly as it renders its own rows' icons
+ * (1.3.0). It exists because without it the six UO rows would have extracted as
+ * the only text-only entries in a sidebar where every other row has a glyph,
+ * which reads as breakage rather than as a design. Core does not supply a
+ * fallback: a module that omits it gets no icon, the same as a core row that
+ * omits it, and inventing one would be core making a presentation choice for
+ * content it knows nothing about. Note that `icon` is already among the fields
+ * an override may not touch (lib/navOverrides.js) — the concept predates a
+ * module being able to supply one.
+ *
  * @param {string} id
- * @param {{area: string, items: Array<{label, to, group?, order?, roles?, feature?}>}} spec
+ * @param {{area: string, items: Array<{label, to, group?, order?, roles?, feature?, icon?}>}} spec
  */
 export function registerNav(id, spec) {
   const { area, items } = spec || {}
@@ -92,10 +102,10 @@ export function registerNav(id, spec) {
 /**
  * The hook that answers "which of this module's features may this viewer see".
  *
- * Core keeps a generic flag context and owns none of the semantics — `uo` fills
- * its namespace with today's `useShardFeatures` (MODULE_SYSTEM.md §1.5). With no
- * module installed the nav filter is a correct no-op, because no core nav item
- * carries a `feature` today.
+ * Core keeps a generic flag context and owns none of the semantics
+ * (MODULE_SYSTEM.md §1.5). With no module installed the nav filter is a correct
+ * no-op, because no core nav item carries a `feature` — which has been literally
+ * true since Phase 3 slice 3 took the nine shard-gated rows out.
  */
 export function registerFeatureProvider(id, namespace, hook) {
   providers.set(namespace, { id, hook })
diff --git a/client/src/modules/shared.js b/client/src/modules/shared.js
index 03a4ad7..9d2a1df 100644
--- a/client/src/modules/shared.js
+++ b/client/src/modules/shared.js
@@ -37,7 +37,7 @@ import { Loading, ErrorState, EmptyState } from '../components/PageState.jsx'
 import { useAsync } from '../lib/useAsync.js'
 import { useAuth } from '../contexts/AuthContext.jsx'
 import { useSite } from '../contexts/SiteContext.jsx'
-import { request, ApiError } from '../api/client.js'
+import { request, ApiError, BASE } from '../api/client.js'
 
 // The UI kit is CURATED AND CLOSED (§3.4), not a re-export of components/. These
 // seven are what the smallest UO page already needs beyond React and the router:
@@ -66,11 +66,16 @@ const ui = {
   useSite,
 }
 
-// The request PRIMITIVE, not the `api` object (§3.5). `api.atlas` and `api.shard`
-// are module bindings that only still live in core's client because Phase 3 has
-// not moved them; a module builds its own namespace over `request` and owns the
-// paths it calls — which is right, because it owns the routes at the other end.
-const api = { request, ApiError }
+// The request PRIMITIVE, not the `api` object (§3.5): a module builds its own
+// namespace over `request` and owns the paths it calls, which is right, because
+// it owns the routes at the other end.
+//
+// `BASE` was in §3.5 from the start and missing from this object until slice 3,
+// which is when something first needed it. `request` is fetch-only, so an
+// EventSource — the shard's live feed is two of them — has to build its own URL,
+// and the alternative is a module hardcoding `/api/v1`: an assertion about where
+// core mounts its API that core has never promised to keep.
+const api = { request, ApiError, BASE }
 
 /**
  * Publish `window.__rg`. Called by main.jsx before it renders, and before any
diff --git a/client/src/modules/version.js b/client/src/modules/version.js
index 466b4c9..9150dc2 100644
--- a/client/src/modules/version.js
+++ b/client/src/modules/version.js
@@ -11,6 +11,12 @@
 // that the two files can drift, so a test asserts they agree
 // (client/test/moduleRegistry.test.js) rather than trusting a bump to remember
 // both.
+// 1.3.0 — three additions, all from Phase 3 slice 3 needing them: a nav item may
+// carry an `icon` component (§3.3), core declares a third slot
+// `player.invite.accepted` (§3.7), and `window.__rg.api` gained `BASE`, which
+// §3.5 always documented and shared.js never published. Additive throughout: a
+// module written against 1.2.0 is unaffected. The server half is untouched and
+// bumps anyway, for the reason below.
 // 1.2.0 — `registry` gained `registerExtension` and core gained extension slots
 // (MODULE_API.md §3.7). The first change to window.__rg since 1.0.0, and an
 // addition: a module that never fills a slot is unaffected. The server half is
@@ -20,4 +26,4 @@
 // but the two halves state ONE version: a module declares a single coreApi range
 // and is served one chunk, so a client that claimed 1.0.0 while the server
 // answered 1.1.0 would be two answers to one question.
-export const MODULE_API_VERSION = '1.2.0'
+export const MODULE_API_VERSION = '1.3.0'
diff --git a/client/src/routes/player/AcceptInvite.jsx b/client/src/routes/player/AcceptInvite.jsx
index 84a82db..3271bb1 100644
--- a/client/src/routes/player/AcceptInvite.jsx
+++ b/client/src/routes/player/AcceptInvite.jsx
@@ -1,14 +1,22 @@
-import { useEffect, useState } from 'react'
+import { useCallback, useEffect, useState } from 'react'
 import { Link, useNavigate, useParams } from 'react-router-dom'
 import { useAuth } from '../../contexts/AuthContext.jsx'
 import { api } from '../../api/client.js'
 import PlayerShell, { honeypotStyle } from './PlayerShell.jsx'
-import CreateGameAccountForm from '../../components/CreateGameAccountForm.jsx'
+import Slot from '../../modules/Slot.jsx'
+import { extensionFor } from '../../modules/registry.js'
 
 // Public, token-gated invite acceptance (/invite/:token). Validates the invite,
 // lets the invitee set a username + password (their email + role are pre-assigned),
-// creates the account at that role and logs them in. For a player invite it then
-// offers the built-in "create game account" step before sending them to the portal.
+// creates the account at that role and logs them in.
+//
+// For a PLAYER invite there may then be one more step, supplied by an installed
+// module through the `player.invite.accepted` slot: core rendered a UO
+// game-account form here itself until Phase 3 slice 3, reading a
+// `gameAccountSignup` flag out of its own settings and posting to a shard route.
+// Neither of those is core's. What core keeps is the shell, the skip control and
+// the destination; whether there is a step at all is the module's call, made
+// from data core does not have.
 export default function AcceptInvite() {
   const { token } = useParams()
   const navigate = useNavigate()
@@ -16,7 +24,6 @@ export default function AcceptInvite() {
 
   const [invite, setInvite] = useState(null) // fields email and role
   const [loadErr, setLoadErr] = useState('')
-  const [signupOk, setSignupOk] = useState(false)
 
   const [username, setUsername] = useState('')
   const [password, setPassword] = useState('')
@@ -30,14 +37,20 @@ export default function AcceptInvite() {
     api.getInvite(token)
       .then((iv) => active && setInvite(iv))
       .catch((err) => active && setLoadErr(err.status === 404 ? 'This invitation is invalid or has expired.' : 'Could not load this invitation.'))
-    api.publicSettings()
-      .then((s) => active && setSignupOk(Boolean(s?.gameAccountSignup)))
-      .catch(() => {})
     return () => { active = false }
   }, [token])
 
   const dest = invite && invite.role === 'player' ? '/player' : '/admin'
 
+  // Whether anything is installed that wants the post-acceptance step. Read
+  // rather than rendered blind because it decides a NAVIGATION, not just what
+  // appears: with nothing filled there is no screen to show, so the invitee goes
+  // straight to their destination. This is the one legitimate reason to ask
+  // whether a slot is filled — the answer changes control flow, not decoration
+  // (decoration goes inside `<Slot wrap>`, which is why `hasExtension` is gone).
+  const hasNextStep = Boolean(extensionFor('player.invite.accepted'))
+  const finish = useCallback(() => navigate('/player', { replace: true }), [navigate])
+
   async function onSubmit(e) {
     e.preventDefault()
     setError('')
@@ -48,8 +61,9 @@ export default function AcceptInvite() {
       await api.acceptInvite(token, username.trim(), password, { company })
       await refresh() // pull the freshly-issued session into context
       setAccepted(true)
-      // Staff invites are web-only — no game step; go straight in.
-      if (!(invite.role === 'player' && signupOk)) navigate(dest, { replace: true })
+      // Staff invites go straight in, and so does a player invite when nothing
+      // is installed that has a step to offer.
+      if (!(invite.role === 'player' && hasNextStep)) navigate(dest, { replace: true })
     } catch (err) {
       if (err.status === 409) setError('That username is already taken, or the invite was already used.')
       else if (err.status === 404) setError('This invitation is invalid or has expired.')
@@ -78,19 +92,22 @@ export default function AcceptInvite() {
     )
   }
 
-  // ── Accepted: optional game-account step (player invites) ──────────────────
+  // ── Accepted: a module's optional next step (player invites) ───────────────
+  //
+  // Only reachable when the slot is filled — `onSubmit` navigates away otherwise
+  // — so there is no empty-shell case to guard here.
+  //
+  // The subtitle is core's and says nothing about what the step is: naming it
+  // would be core describing content it does not own, and the wrong description
+  // is worse than a general one. "Skip" stays core's too, because where it goes
+  // is core's decision, and it is rendered outside the slot deliberately — an
+  // extension that throws must not take the way out with it.
   if (accepted) {
     return (
-      <PlayerShell subtitle="Set up your game account">
-        <p className="sans" style={{ marginTop: 0, color: 'var(--muted)', fontSize: '0.9rem', lineHeight: 1.6 }}>
-          Your account is ready. Create a game account now to play, or skip and do it later from your portal.
-        </p>
-        <CreateGameAccountForm
-          submit={api.player.shard.createAccount}
-          onCreated={() => navigate('/player', { replace: true })}
-        />
+      <PlayerShell subtitle="One more step">
+        <Slot name="player.invite.accepted" onDone={finish} />
         <p className="sans" style={{ textAlign: 'center', margin: '18px 0 0' }}>
-          <button type="button" onClick={() => navigate('/player', { replace: true })} className="btn" style={{ background: 'none', border: 'none', color: 'var(--accent)', cursor: 'pointer' }}>
+          <button type="button" onClick={finish} className="btn" style={{ background: 'none', border: 'none', color: 'var(--accent)', cursor: 'pointer' }}>
             Skip for now →
           </button>
         </p>
diff --git a/client/test/moduleNav.test.js b/client/test/moduleNav.test.js
index 011dd02..c1209db 100644
--- a/client/test/moduleNav.test.js
+++ b/client/test/moduleNav.test.js
@@ -132,6 +132,36 @@ test('a module row carries its moduleId through, which is how the gate finds it'
   assert.equal(row.feature, 'atlas')
 })
 
+test('a module row carries its icon through, and an override cannot touch it', () => {
+  // 1.3.0. Without an `icon` the six extracted UO rows would have been the only
+  // text-only entries in a sidebar where every other row has a glyph. Core
+  // renders whatever component the row carries and supplies no fallback — an
+  // invented one would be core making a presentation choice for content it knows
+  // nothing about.
+  const Glyph = () => null
+  registerNav('uo', {
+    area: 'admin',
+    items: [{ label: 'Shard', to: '/admin/uo/link', group: 'System', icon: Glyph }],
+  })
+  const merged = withModuleNav(ADMIN, 'admin')
+  const row = merged.find((g) => g.title === 'System').items.find((i) => i.to === '/admin/uo/link')
+  assert.equal(row.icon, Glyph)
+
+  // `icon` was already on navOverrides' list of fields an override may not
+  // touch, from long before a module could supply one. It still is.
+  const overridden = applyNavOverrides(merged, { '/admin/uo/link': { label: 'Renamed', icon: 'nope' } })
+  const after = overridden.find((g) => g.title === 'System').items.find((i) => i.to === '/admin/uo/link')
+  assert.equal(after.label, 'Renamed')
+  assert.equal(after.icon, Glyph)
+})
+
+test('a row with no icon simply has none, the same as a core row with none', () => {
+  registerNav('uo', { area: 'player', items: [{ label: 'Characters', to: '/player/uo/characters', order: 0 }] })
+  const row = withModuleNav([{ to: '/account', label: 'Account' }], 'player')[0]
+  assert.equal(row.to, '/player/uo/characters')
+  assert.equal(row.icon, undefined)
+})
+
 test('areas do not leak into one another', () => {
   registerNav('uo', { area: 'admin', items: [{ label: 'Shard', to: '/admin/uo/link', group: 'System' }] })
   assert.equal(withModuleNav(PUBLIC, 'public'), PUBLIC)
diff --git a/server/src/modules/version.js b/server/src/modules/version.js
index 24b34bc..031ae8d 100644
--- a/server/src/modules/version.js
+++ b/server/src/modules/version.js
@@ -9,6 +9,11 @@
 // Deliberately separate from PROTOCOL_VERSION (which versions the shard wire and
 // has nothing to say about a website module) and from any module's own version.
 
+// 1.3.0 — three CLIENT additions from Phase 3 slice 3: a nav item may carry an
+// `icon`, core declares a `player.invite.accepted` slot, and `window.__rg.api`
+// gained `BASE` (which §3.5 always specified and shared.js never published).
+// Nothing on the server changed; this file bumps for the reason below.
+//
 // 1.2.0 — the CLIENT registry gained `registerExtension` and core gained client
 // extension slots (MODULE_API.md §3.7): the twin of this half's declareSlot /
 // registerExtension, for module content inside a core *page* rather than under a
@@ -21,6 +26,6 @@
 // an admin action a module performs belongs in core's one audit log, the
 // extension slot needs the user its prefix names, and §2.7 forbids a module
 // reading core's `APP_BASE_URL` for itself. Additions only, so minor.
-const MODULE_API_VERSION = '1.2.0'
+const MODULE_API_VERSION = '1.3.0'
 
 module.exports = { MODULE_API_VERSION }