diff --git a/docs/systems/dm-system.md b/docs/systems/dm-system.md
index 710d015b..fc53b79a 100644
--- a/docs/systems/dm-system.md
+++ b/docs/systems/dm-system.md
@@ -284,6 +284,7 @@ If a `member_add` federation event arrives for a soft-deleted channel (non-null
 **Request:** `{ content?: string, attachments?: string[], replyToId?: string }`
 
 **Cross-instance access:** Federated users (those with `homeInstance` set) can send messages on any DM channel where they are a member, regardless of which instance serves the request. The `requireLocalUser` gate that previously blocked federated users from DM write endpoints has been removed. DM calls work across federated instances. The caller's instance hosts the LiveKit room; remote clients connect directly. Call signaling is relayed to all active federation peers via synchronous HTTP POST (not the outbox worker). Relay failures at any call state transition emit `dm_call_undeliverable { phase, terminal, failures }` to the originator — see `docs/systems/voice.md` for the full call state machine and failure surface.
+- Federated call-start to a remote instance with no reachable recipient surfaces as `dm_call_undeliverable` with reason `no_recipient` — see `voice.md` for the full failure-surface table.
 
 **Validation:**
 - Caller must be a member (`isDmMember`)
diff --git a/docs/systems/federation.md b/docs/systems/federation.md
index 8e69dc62..9f817d57 100644
--- a/docs/systems/federation.md
+++ b/docs/systems/federation.md
@@ -529,10 +529,42 @@ interface FederationRelayResponse {
     messageId: string;
     reason: string;                // e.g., 'duplicate', 'unknown_message', 'missing_participants'
   }>;
+  undeliverable?: Array<{          // optional — omitted when empty; call-signaling only
+    messageId: string;
+    reason: string;                // e.g., 'no_recipient'
+  }>;
   maxUploadSize: number;           // This instance's max upload size in bytes
 }
 ```
 
+#### `undeliverable` bucket (call-signaling only)
+
+In addition to `accepted` and `rejected`, the relay response may include an
+optional `undeliverable: Array<{messageId, reason}>`. Three-way classification,
+non-overlapping: each messageId appears in exactly one of the three arrays.
+
+| Bucket | Meaning | Retry? |
+|---|---|---|
+| `accepted` | Processed cleanly, ≥1 recipient reached. | No |
+| `rejected` | Refused at data/protocol layer (schema, attribution, channel-not-found, etc.). | Terminal. |
+| `undeliverable` | Processed cleanly, zero recipients reachable. | No — call-signaling specific. |
+
+Currently used only for `dm_call_start`:
+- **Path A** (local DM exists): if no local non-caller member has an active WS
+  connection, the event is pushed to `undeliverable` with reason `no_recipient`
+  instead of being silently accepted. No `FederatedCallEntry` is created.
+- **Path B** (no local DM): the zero-participant-match early return pushes to
+  `undeliverable` rather than `accepted`.
+
+Other event types (messages, reactions, friend events, profile updates, etc.)
+keep existing semantics — a message to an offline user is still `accepted`, since
+messages persist and re-deliver on reconnect.
+
+The field is optional on the wire. Old peers omit it; new peers include it only
+when non-empty. Caller-side `sendCallRelay` parses the field (defaulting to an
+empty array when missing), so upgrade skew is a no-op until both sides are on
+new code.
+
 ### Inbound Relay Dispatch (`POST /api/federation/relay`)
 
 Body limit: 10 MB. Max 50 events per batch. Rate-limited to 90 requests/min per peer (sliding window, keyed by `peer.origin`). Returns 429 when exceeded. Raised from 30 after FED-009 reduced the outbox worker interval from 10s to 1s — a busy sender can now hit 60 req/min during sustained traffic.
diff --git a/docs/systems/voice.md b/docs/systems/voice.md
index 7d396a7c..8ee143b5 100644
--- a/docs/systems/voice.md
+++ b/docs/systems/voice.md
@@ -71,6 +71,7 @@ All `dm_call_*` signaling events (`start`, `accept`, `reject`, `end`) are relaye
 | `reject` | false | Rejector's relay to host failed OR host's fan-out after a local reject failed; state already cleared. | No state change; info toast. |
 | `end` | false | Ender's relay to host failed OR host's fan-out after a local end failed; state already cleared. | No state change; info toast. |
 | `host_unreachable` | true | A FederatedCallEntry's `federatedCallHost` peer transitions out of `active`, OR the 30s sentinel detects a non-active host for an existing entry. | Clear `activeDmCall` + `incomingCall`, disconnect LK, warning toast (*"Call ended — {label} became unreachable."*). |
+| `no_recipient` | true | Remote returned 200 but had no reachable recipient (Path A: all members offline; Path B: zero participant matches). Caller fast-fails within the relay round-trip; ring room destroyed. | Clear `outgoingCall`, disconnect LK, warning toast (*"{peerLabel} couldn't ring anyone."*). Folds into multi-failure info copy when not the sole failure. |
 
 **Accept-rollback semantics.** `handleDmCallAccept` Path 2 transitions the `FederatedCallEntry` to active and broadcasts `dm_call_accepted` optimistically so the acceptor's UI flips immediately. If the B→host relay fails, the server clears the entry, fans `dm_call_undeliverable { phase: 'accept', terminal: true }` out to all ringed users on B (via `sendToFederatedCallUsers`), and the client tears its call state back down.
 
diff --git a/docs/systems/websocket.md b/docs/systems/websocket.md
index 0d40582f..b3e7ca13 100644
--- a/docs/systems/websocket.md
+++ b/docs/systems/websocket.md
@@ -170,7 +170,7 @@ reason: `'displaced'` (new tab) | `'session_closed'`
 | `dm_call_accepted` | dmChannelId?, federatedCallId? | DM members |
 | `dm_call_rejected` | dmChannelId?, federatedCallId? | DM members |
 | `dm_call_ended` | dmChannelId?, federatedCallId? | DM members |
-| `dm_call_undeliverable` | Sent to the originator when a call relay (start / accept / reject / end) to one or more peers fails. Includes `phase: 'start' \| 'accept' \| 'reject' \| 'end'` identifying the action; `failures[]` enumerates failed peers with a `reason` (`peer_rejected` / `peer_awaiting_approval` / `peer_transient_failure` / `livekit_unavailable`). `terminal: true` means local call state should be (or has been) torn down; `terminal: false` is informational. See `docs/systems/voice.md` for the full phase × terminal matrix. | originator (caller / acceptor / rejector / ender) |
+| `dm_call_undeliverable` | Sent to the originator when a call relay (start / accept / reject / end) to one or more peers fails. Includes `phase: 'start' \| 'accept' \| 'reject' \| 'end'` identifying the action; `failures[]` enumerates failed peers with a `reason` (`peer_rejected` / `peer_awaiting_approval` / `peer_transient_failure` / `livekit_unavailable` / `no_recipient`). `terminal: true` means local call state should be (or has been) torn down; `terminal: false` is informational. See `docs/systems/voice.md` for the full phase × terminal matrix. | originator (caller / acceptor / rejector / ender) |
 
 ### Social
 | type | fields | scope |