Presence & status model + endpoints (Redis heartbeat) -- recovered

Real feature built in August, checkpointed mid-task by the old fleet's
graceful-shutdown mechanism, then orphaned when checkpoint_focus_
feature_ids scoped the fleet away from M8 and it was never resumed --
the PM task had shown done/reviewed for a month with zero actual code on
main. Ported forward from the 225-commits-stale branch onto current
main's conventions (DbDep pattern, migration rebased onto 0005's head),
fixing two real bugs found in the process: the original PresenceStore's
explicit connect() was never called from anywhere (would 500 on every
real request), and a naive fix hits 'event loop is closed' under this
repo's TestClient(app) pattern -- fixed with lazy per-event-loop client
caching, same class of issue already documented for asyncpg in
conftest.py. conftest.py extended to also bring up ephemeral Redis
symmetrically with its existing Postgres bootstrap.

144/144 backend tests independently re-run via the fully automated
hermetic path (zero manual setup, matching what a fresh contributor
gets), plus a manual end-to-end heartbeat/TTL-expiry/batch-query check
against real Postgres+Redis.

backend/alembic/versions/0006_add_presence_columns.py

@@ -0,0 +1,33 @@
+"""add presence columns to users
+
+Revision ID: 0006_add_presence_columns
+Revises: 0005_add_world_content_tables
+Create Date: 2026-09-14 00:00:00.000000
+
+Ported forward from the orphaned agent/dd2a126b57a1 branch (originally
+2a9b9873244a_add_presence_columns.py, down_revision 0001_initial) and
+rebased onto the current migration head. See app/services/presence.py /
+app/routers/presence.py.
+"""
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+# revision identifiers, used by Alembic.
+revision: str = "0006_add_presence_columns"
+down_revision: Union[str, None] = "0005_add_world_content_tables"
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    """Add last_seen_at and status_message columns to users table."""
+    op.add_column("users", sa.Column("last_seen_at", sa.DateTime(timezone=True), nullable=True))
+    op.add_column("users", sa.Column("status_message", sa.Text(), nullable=True))
+
+
+def downgrade() -> None:
+    """Remove presence columns from users table."""
+    op.drop_column("users", "status_message")
+    op.drop_column("users", "last_seen_at")

backend/app/main.py

@@ -6,7 +6,7 @@ from fastapi import FastAPI
 from fastapi.middleware.cors import CORSMiddleware
 
 from app.config import settings
-from app.routers import auth, users, worlds, assets, avatar, world_content, internal
+from app.routers import auth, users, worlds, assets, avatar, world_content, internal, presence
 
 
 from app.services.game_servers import supervisor
@@ -44,6 +44,7 @@ app.include_router(world_content.router, prefix="/api/worlds", tags=["world-cont
 app.include_router(assets.router, prefix="/api/assets", tags=["assets"])
 app.include_router(avatar.router, prefix="/api/avatar", tags=["avatar"])
 app.include_router(internal.router, prefix="/api/internal", tags=["internal"])
+app.include_router(presence.router, prefix="/api/presence", tags=["presence"])
 
 
 @app.get("/api/health")

backend/app/models/user.py

@@ -1,7 +1,7 @@
 from datetime import date
 from typing import Optional
 
-from sqlalchemy import Column, Integer, String, Boolean, DateTime, Date
+from sqlalchemy import Column, Integer, String, Boolean, DateTime, Date, Text
 from sqlalchemy.orm import relationship
 from sqlalchemy.sql import func
 
@@ -20,6 +20,10 @@ class User(Base):
     created_at = Column(DateTime(timezone=True), server_default=func.now())
     updated_at = Column(DateTime(timezone=True), onupdate=func.now())
 
+    # Presence / status (backend/app/services/presence.py, app/routers/presence.py)
+    last_seen_at = Column(DateTime(timezone=True), nullable=True)
+    status_message = Column(Text, nullable=True)
+
     sessions = relationship("Session", back_populates="user", cascade="all, delete-orphan")
     refresh_tokens = relationship("RefreshToken", back_populates="user", cascade="all, delete-orphan")
     worlds = relationship("World", back_populates="owner")

backend/app/routers/presence.py

@@ -0,0 +1,146 @@
+"""Presence & status router.
+
+Endpoints:
+- POST  /api/presence/heartbeat  -- record a heartbeat (sets Redis TTL key)
+- GET   /api/presence/presence   -- query presence for a list of user ids
+- PATCH /api/presence/status     -- update the authenticated user's status message
+"""
+
+from datetime import datetime, timezone
+from typing import Annotated
+
+from fastapi import APIRouter, HTTPException, status, Depends
+from pydantic import BaseModel, Field, field_validator
+from sqlalchemy import select
+from sqlalchemy.ext.asyncio import AsyncSession
+
+from app.database import get_db
+from app.dependencies import CurrentUser
+from app.models.user import User
+from app.services.presence import presence_store
+
+router = APIRouter()
+
+DbDep = Annotated[AsyncSession, Depends(get_db)]
+
+# ──────────────────────────── Pydantic schemas ────────────────────────────────
+
+_STATUS_MAX_LEN = 140
+
+
+class HeartbeatRequest(BaseModel):
+    """Empty body -- the caller's identity comes from the JWT."""
+
+
+class PresenceUser(BaseModel):
+    user_id: int
+    is_online: bool
+    last_seen_at: datetime | None
+    status_message: str | None
+
+
+class PresenceResponse(BaseModel):
+    users: list[PresenceUser]
+
+
+class StatusUpdate(BaseModel):
+    status_message: str | None = Field(default=None, max_length=_STATUS_MAX_LEN)
+
+    @field_validator("status_message")
+    @classmethod
+    def strip_message(cls, v: str | None) -> str | None:
+        if v is not None:
+            v = v.strip()
+        return v
+
+
+# ──────────────────────────────── Endpoints ───────────────────────────────────
+
+
+@router.post("/heartbeat", response_model=dict)
+async def heartbeat(
+    _req: HeartbeatRequest,
+    current_user: CurrentUser,
+    db: DbDep,
+):
+    """Record a heartbeat for the authenticated user.
+
+    Sets a Redis key with a TTL so the server can derive online/offline
+    status without trusting client claims. Also persists last_seen_at to
+    the database.
+    """
+    # Update Redis TTL (presence heartbeat)
+    await presence_store.heartbeat(current_user.id)
+
+    # Persist last_seen_at to the database
+    now = datetime.now(timezone.utc)
+    result = await db.execute(select(User).where(User.id == current_user.id))
+    user = result.scalar_one_or_none()
+    if user:
+        user.last_seen_at = now
+        await db.commit()
+
+    return {"status": "ok", "user_id": current_user.id}
+
+
+@router.get("/presence", response_model=PresenceResponse)
+async def get_presence(
+    user_ids: str,
+    current_user: CurrentUser,
+    db: DbDep,
+):
+    """Return presence info for a comma-separated list of user ids.
+
+    Online/offline is derived from the Redis heartbeat TTL.
+    last_seen_at and status_message come from the database.
+    """
+    ids = [int(uid.strip()) for uid in user_ids.split(",") if uid.strip()]
+    if not ids:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail="user_ids must contain at least one integer",
+        )
+
+    # Check Redis for online status (single round trip via MGET).
+    online_ids = await presence_store.is_online_batch(ids)
+
+    # Fetch DB records for last_seen_at and status_message
+    result = await db.execute(select(User).where(User.id.in_(ids)))
+    users = {u.id: u for u in result.scalars().all()}
+
+    presence_users = []
+    for uid in ids:
+        user = users.get(uid)
+        if user is None:
+            continue
+        presence_users.append(
+            PresenceUser(
+                user_id=uid,
+                is_online=uid in online_ids,
+                last_seen_at=user.last_seen_at,
+                status_message=user.status_message,
+            )
+        )
+
+    return PresenceResponse(users=presence_users)
+
+
+@router.patch("/status", response_model=dict)
+async def update_status(
+    body: StatusUpdate,
+    current_user: CurrentUser,
+    db: DbDep,
+):
+    """Update the authenticated user's status message (length-limited)."""
+    result = await db.execute(select(User).where(User.id == current_user.id))
+    user = result.scalar_one_or_none()
+    if not user:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail="User not found",
+        )
+
+    user.status_message = body.status_message
+    await db.commit()
+
+    return {"status": "ok", "status_message": body.status_message}

backend/app/services/presence.py

@@ -0,0 +1,145 @@
+"""Redis-backed presence store.
+
+Presence is determined by a heartbeat TTL key in Redis:
+- heartbeat(user_id) sets a TTL key per user (default 60 s).
+- is_online(user_id) / is_online_batch(user_ids) check whether the key(s)
+  exist.
+
+status_message and last_seen_at are persisted in the ``users`` table by the
+router (app/routers/presence.py); this store only owns the Redis side.
+
+The Redis client is created lazily, on first actual use, rather than
+requiring an explicit ``connect()`` call during app startup. This matters
+because FastAPI's lifespan handler does not run under the plain
+``TestClient(app)`` instantiation this codebase's tests use (no
+``with`` block), so anything that only connected during lifespan startup
+would never connect in tests. Lazy connection means the store works
+identically whether the app was started for real or is being driven by
+pytest.
+
+The client is also cached per-*event loop* rather than globally: the same
+plain ``TestClient(app)`` pattern can dispatch different requests onto
+different event loops (see the NullPool comment on ``_override_db`` in
+tests/conftest.py for the analogous asyncpg problem), and an
+``aioredis.Redis`` connection, like an asyncpg connection, breaks with
+"Event loop is closed" if reused across loops. In the real running server
+there is exactly one event loop for the process lifetime, so this
+degrades to a single cached, pooled connection there.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import logging
+
+import redis.asyncio as aioredis
+
+from app.config import settings
+
+logger = logging.getLogger(__name__)
+
+# Key prefix for heartbeat entries.
+_PRESENCE_KEY_PREFIX = "presence:heartbeat:"
+
+# Default heartbeat TTL in seconds (1 minute).
+# Clients should heartbeat at least once per interval.
+_DEFAULT_HEARTBEAT_TTL = 60
+
+
+class PresenceStore:
+    """Redis-backed presence store with async, lazily-connected operations."""
+
+    def __init__(self, redis_url: str | None = None, heartbeat_ttl: int = _DEFAULT_HEARTBEAT_TTL):
+        # redis_url is resolved from settings at first use (not at
+        # construction time) so tests that override settings.REDIS_URL
+        # before the first call still take effect -- this module is
+        # imported (and the module-level singleton below constructed)
+        # well before any such override runs.
+        self._redis_url = redis_url
+        self._heartbeat_ttl = heartbeat_ttl
+        # Keyed by id(running loop) -- see the module docstring.
+        self._clients: dict[int, aioredis.Redis] = {}
+        self._connect_lock = asyncio.Lock()
+
+    # ------------------------------------------------------------------
+    # Lifecycle
+    # ------------------------------------------------------------------
+
+    async def _ensure_client(self) -> aioredis.Redis:
+        """Return a connected client for the current event loop, connecting
+        on first use per loop."""
+        loop_key = id(asyncio.get_running_loop())
+        client = self._clients.get(loop_key)
+        if client is None:
+            async with self._connect_lock:
+                client = self._clients.get(loop_key)
+                if client is None:
+                    url = self._redis_url or settings.REDIS_URL
+                    client = aioredis.from_url(
+                        url,
+                        encoding="utf-8",
+                        decode_responses=True,
+                    )
+                    try:
+                        await client.ping()
+                    except Exception:
+                        logger.warning("Redis ping failed for %s", url)
+                        raise
+                    self._clients[loop_key] = client
+        return client
+
+    async def close(self) -> None:
+        """Close the Redis connection for the current event loop, if one was
+        opened."""
+        loop_key = id(asyncio.get_running_loop())
+        client = self._clients.pop(loop_key, None)
+        if client is not None:
+            await client.aclose()
+
+    # ------------------------------------------------------------------
+    # Heartbeat
+    # ------------------------------------------------------------------
+
+    async def heartbeat(self, user_id: int) -> None:
+        """Record a heartbeat for *user_id*.
+
+        Sets a TTL key so that the user is considered online as long as
+        heartbeats arrive at least once per heartbeat_ttl seconds.
+        """
+        client = await self._ensure_client()
+        await client.set(self._heartbeat_key(user_id), "1", ex=self._heartbeat_ttl)
+
+    # ------------------------------------------------------------------
+    # Online checks
+    # ------------------------------------------------------------------
+
+    async def is_online(self, user_id: int) -> bool:
+        """Return True if the user has a valid heartbeat key."""
+        client = await self._ensure_client()
+        exists = await client.exists(self._heartbeat_key(user_id))
+        return bool(exists)
+
+    async def is_online_batch(self, user_ids: list[int]) -> set[int]:
+        """Return the subset of user_ids that currently have a valid
+        heartbeat key. Uses MGET (one round trip) rather than one EXISTS
+        per user.
+        """
+        if not user_ids:
+            return set()
+        client = await self._ensure_client()
+        keys = [self._heartbeat_key(uid) for uid in user_ids]
+        values = await client.mget(keys)
+        return {uid for uid, v in zip(user_ids, values) if v is not None}
+
+    # ------------------------------------------------------------------
+    # Helpers
+    # ------------------------------------------------------------------
+
+    @staticmethod
+    def _heartbeat_key(user_id: int) -> str:
+        return f"{_PRESENCE_KEY_PREFIX}{user_id}"
+
+
+# Singleton instance shared by the router, so all requests share the same
+# connection pool.
+presence_store = PresenceStore()

backend/tests/conftest.py

@@ -1,15 +1,18 @@
-"""Shared pytest fixtures + hermetic-DB bootstrap.
+"""Shared pytest fixtures + hermetic-DB/Redis bootstrap.
 
 The integration tests hit a real Postgres (Postgres-specific types: JSONB,
-native ENUMs). To make ``cd backend && pytest`` work from a clean checkout
-with no manual setup, ``pytest_configure`` below brings up an ephemeral
-Postgres via docker compose and runs the migrations IF ``DATABASE_URL`` is
-not already pointing at a reachable database. When it is (dev loop, or the
-checkpoint runner which stands the DB up itself), that path is skipped and
-the existing database is reused.
+native ENUMs) and, for presence tests, a real Redis (TTL semantics aren't
+worth faking). To make ``cd backend && pytest`` work from a clean checkout
+with no manual setup, ``pytest_configure`` below brings up ephemeral
+Postgres/Redis containers via docker compose and runs the migrations IF
+``DATABASE_URL``/``REDIS_URL`` aren't already pointing at something
+reachable. When they are (dev loop, or the checkpoint runner which stands
+these up itself), that path is skipped and the existing services are
+reused.
 
-If Docker isn't available and no DB is reachable, the whole DB-backed suite
-is skipped with a clear message rather than erroring out 60+ times.
+If Docker isn't available and nothing is reachable, the whole
+DB/Redis-backed suite is skipped with a clear message rather than erroring
+out dozens of times.
 """
 
 import asyncio
@@ -44,8 +47,9 @@ TEST_TABLES = (
 BACKEND_DIR = Path(__file__).resolve().parents[1]
 _COMPOSE_PROJECT = "agp-pytest"
 _EPHEMERAL_PG_PORT = 25432
+_EPHEMERAL_REDIS_PORT = 26379
 
-# Set by _ensure_database() when it starts a container it owns; torn down in
+# Set by pytest_configure() when it starts container(s) it owns; torn down in
 # pytest_unconfigure.
 _started_container = False
 _db_unavailable_reason: str | None = None
@@ -64,6 +68,11 @@ def _db_reachable(url: str) -> bool:
     return _port_open(parts.hostname or "localhost", parts.port or 5432)
 
 
+def _redis_reachable(url: str) -> bool:
+    parts = urlsplit(url)
+    return _port_open(parts.hostname or "localhost", parts.port or 6379)
+
+
 def _run(cmd: list[str], **kw) -> subprocess.CompletedProcess:
     return subprocess.run(cmd, cwd=BACKEND_DIR, capture_output=True, text=True, **kw)
 
@@ -71,14 +80,17 @@ def _run(cmd: list[str], **kw) -> subprocess.CompletedProcess:
 def pytest_configure(config):
     global _started_container, _db_unavailable_reason
 
-    if _db_reachable(settings.DATABASE_URL):
-        return  # dev loop / checkpoint runner already provided one
+    need_db = not _db_reachable(settings.DATABASE_URL)
+    need_redis = not _redis_reachable(settings.REDIS_URL)
+
+    if not need_db and not need_redis:
+        return  # dev loop / checkpoint runner already provided both
 
     if not _run(["docker", "--version"]).returncode == 0:
         _db_unavailable_reason = (
-            "no reachable DATABASE_URL and docker is unavailable -- run the "
-            "checkpoint suite (scripts/run_checkpoint_tests.py) or start Postgres "
-            "manually"
+            "no reachable DATABASE_URL/REDIS_URL and docker is unavailable -- run "
+            "the checkpoint suite (scripts/run_checkpoint_tests.py) or start "
+            "Postgres/Redis manually"
         )
         return
 
@@ -86,33 +98,55 @@ def pytest_configure(config):
     env["COMPOSE_PROJECT_NAME"] = _COMPOSE_PROJECT
     env["POSTGRES_HOST_PORT"] = str(_EPHEMERAL_PG_PORT)
     env["POSTGRES_PORT"] = str(_EPHEMERAL_PG_PORT)
-    up = _run(["docker", "compose", "up", "-d", "postgres"], env=env)
+    env["REDIS_HOST_PORT"] = str(_EPHEMERAL_REDIS_PORT)
+
+    services = [s for s, need in (("postgres", need_db), ("redis", need_redis)) if need]
+    up = _run(["docker", "compose", "up", "-d", *services], env=env)
     if up.returncode != 0:
-        _db_unavailable_reason = f"docker compose up postgres failed:\n{up.stderr[-800:]}"
+        _db_unavailable_reason = f"docker compose up {' '.join(services)} failed:\n{up.stderr[-800:]}"
         return
     _started_container = True
 
-    for _ in range(30):
-        r = _run(
-            ["docker", "compose", "exec", "-T", "postgres",
-             "pg_isready", "-U", "agp_user", "-d", "agp_db"],
-            env=env,
-        )
-        if r.returncode == 0:
-            break
-        time.sleep(1)
-    else:
-        _db_unavailable_reason = "ephemeral Postgres never became ready"
-        return
+    if need_db:
+        for _ in range(30):
+            r = _run(
+                ["docker", "compose", "exec", "-T", "postgres",
+                 "pg_isready", "-U", "agp_user", "-d", "agp_db"],
+                env=env,
+            )
+            if r.returncode == 0:
+                break
+            time.sleep(1)
+        else:
+            _db_unavailable_reason = "ephemeral Postgres never became ready"
+            return
 
-    url = f"postgresql+asyncpg://agp_user:agp_password@localhost:{_EPHEMERAL_PG_PORT}/agp_db"
-    os.environ["DATABASE_URL"] = url
-    settings.DATABASE_URL = url
+        url = f"postgresql+asyncpg://agp_user:agp_password@localhost:{_EPHEMERAL_PG_PORT}/agp_db"
+        os.environ["DATABASE_URL"] = url
+        settings.DATABASE_URL = url
 
-    mig = _run([sys.executable, "-m", "alembic", "upgrade", "head"],
-               env={**os.environ, "DATABASE_URL": url})
-    if mig.returncode != 0:
-        _db_unavailable_reason = f"alembic upgrade head failed:\n{mig.stderr[-800:]}"
+        mig = _run([sys.executable, "-m", "alembic", "upgrade", "head"],
+                   env={**os.environ, "DATABASE_URL": url})
+        if mig.returncode != 0:
+            _db_unavailable_reason = f"alembic upgrade head failed:\n{mig.stderr[-800:]}"
+            return
+
+    if need_redis:
+        for _ in range(30):
+            r = _run(
+                ["docker", "compose", "exec", "-T", "redis", "redis-cli", "ping"],
+                env=env,
+            )
+            if r.returncode == 0 and "PONG" in r.stdout:
+                break
+            time.sleep(1)
+        else:
+            _db_unavailable_reason = "ephemeral Redis never became ready"
+            return
+
+        redis_url = f"redis://localhost:{_EPHEMERAL_REDIS_PORT}/0"
+        os.environ["REDIS_URL"] = redis_url
+        settings.REDIS_URL = redis_url
 
 
 def pytest_unconfigure(config):
@@ -122,6 +156,7 @@ def pytest_unconfigure(config):
     env["COMPOSE_PROJECT_NAME"] = _COMPOSE_PROJECT
     env["POSTGRES_HOST_PORT"] = str(_EPHEMERAL_PG_PORT)
     env["POSTGRES_PORT"] = str(_EPHEMERAL_PG_PORT)
+    env["REDIS_HOST_PORT"] = str(_EPHEMERAL_REDIS_PORT)
     _run(["docker", "compose", "down", "-v"], env=env)
 
 

backend/tests/test_presence.py

@@ -0,0 +1,221 @@
+"""Integration tests for the presence feature:
+
+- POST  /api/presence/heartbeat
+- GET   /api/presence/presence
+- PATCH /api/presence/status
+
+Hits the real app against the configured Postgres *and* a real Redis
+(conftest.py brings up an ephemeral instance of each when neither is
+already reachable) -- presence is fundamentally "does a TTL key exist in
+Redis", which isn't worth faking with a mock.
+
+Also covers the regression fixed in 47ab492 ("Fix StatusUpdate field
+annotation and is_online batch-call mismatch"): the router used to call
+``PresenceStore.is_online(ids)`` with a *list* of ids and then evaluate
+``uid in online_ids`` against the returned bool, which raised a TypeError
+on the first real multi-id batch call, and ``StatusUpdate.status_message``
+used the invalid annotation ``str | Field(...)``. Both are exercised below.
+"""
+
+import asyncio
+
+import pytest
+from fastapi.testclient import TestClient
+
+from app.main import app
+from app.services.presence import PresenceStore
+
+
+def _register(client: TestClient, username: str) -> dict:
+    resp = client.post(
+        "/api/auth/register",
+        json={
+            "username": username,
+            "email": f"{username}@example.com",
+            "password": "secret123",
+            "date_of_birth": "1990-01-01",
+        },
+    )
+    assert resp.status_code == 201, resp.text
+    token = resp.json()["access_token"]
+    headers = {"Authorization": f"Bearer {token}"}
+    me = client.get("/api/users/me", headers=headers)
+    assert me.status_code == 200, me.text
+    return {"headers": headers, "id": me.json()["id"]}
+
+
+class TestHeartbeat:
+    def test_heartbeat_sets_online(self):
+        client = TestClient(app)
+        user = _register(client, "presence_hb1")
+
+        resp = client.post("/api/presence/heartbeat", json={}, headers=user["headers"])
+        assert resp.status_code == 200, resp.text
+        assert resp.json() == {"status": "ok", "user_id": user["id"]}
+
+        presence = client.get(
+            "/api/presence/presence",
+            params={"user_ids": str(user["id"])},
+            headers=user["headers"],
+        )
+        assert presence.status_code == 200, presence.text
+        body = presence.json()["users"]
+        assert len(body) == 1
+        assert body[0]["user_id"] == user["id"]
+        assert body[0]["is_online"] is True
+        assert body[0]["last_seen_at"] is not None
+
+    def test_requires_auth(self):
+        client = TestClient(app)
+        resp = client.post("/api/presence/heartbeat", json={})
+        assert resp.status_code == 401
+
+    def test_never_seen_user_is_offline_with_null_last_seen(self):
+        client = TestClient(app)
+        user = _register(client, "presence_never_hb")
+
+        presence = client.get(
+            "/api/presence/presence",
+            params={"user_ids": str(user["id"])},
+            headers=user["headers"],
+        )
+        assert presence.status_code == 200
+        body = presence.json()["users"][0]
+        assert body["is_online"] is False
+        assert body["last_seen_at"] is None
+
+
+class TestBatchPresence:
+    """Covers the batch-call path fixed in 47ab492 (is_online_batch via MGET)."""
+
+    def test_batch_query_mixes_online_offline_and_unknown_ids(self):
+        client = TestClient(app)
+        online_user = _register(client, "presence_batch_online")
+        offline_user = _register(client, "presence_batch_offline")
+
+        client.post("/api/presence/heartbeat", json={}, headers=online_user["headers"])
+
+        unknown_id = max(online_user["id"], offline_user["id"]) + 999_999
+        ids = f"{online_user['id']},{offline_user['id']},{unknown_id}"
+
+        resp = client.get(
+            "/api/presence/presence",
+            params={"user_ids": ids},
+            headers=online_user["headers"],
+        )
+        assert resp.status_code == 200, resp.text
+        by_id = {u["user_id"]: u for u in resp.json()["users"]}
+
+        # Unknown id is silently dropped (not in the DB), not a 500 or 404.
+        assert unknown_id not in by_id
+        assert by_id[online_user["id"]]["is_online"] is True
+        assert by_id[offline_user["id"]]["is_online"] is False
+
+    def test_empty_user_ids_is_a_400(self):
+        client = TestClient(app)
+        user = _register(client, "presence_empty_ids")
+        resp = client.get(
+            "/api/presence/presence",
+            params={"user_ids": ""},
+            headers=user["headers"],
+        )
+        assert resp.status_code == 400
+
+
+class TestStatusMessage:
+    def test_update_and_read_back_status_message(self):
+        client = TestClient(app)
+        user = _register(client, "presence_status1")
+
+        resp = client.patch(
+            "/api/presence/status",
+            json={"status_message": "  brb, making coffee  "},
+            headers=user["headers"],
+        )
+        assert resp.status_code == 200, resp.text
+        # The router strips whitespace (StatusUpdate.strip_message).
+        assert resp.json()["status_message"] == "brb, making coffee"
+
+        presence = client.get(
+            "/api/presence/presence",
+            params={"user_ids": str(user["id"])},
+            headers=user["headers"],
+        )
+        assert presence.json()["users"][0]["status_message"] == "brb, making coffee"
+
+    def test_status_message_over_max_length_is_rejected(self):
+        client = TestClient(app)
+        user = _register(client, "presence_status_toolong")
+        resp = client.patch(
+            "/api/presence/status",
+            json={"status_message": "x" * 141},
+            headers=user["headers"],
+        )
+        # StatusUpdate.status_message: str | None = Field(..., max_length=140)
+        # This is exactly the annotation 47ab492 fixed (it used to be the
+        # invalid `str | Field(...)`, which breaks model construction
+        # entirely rather than validating length).
+        assert resp.status_code == 422
+
+    def test_status_message_at_max_length_is_accepted(self):
+        client = TestClient(app)
+        user = _register(client, "presence_status_maxlen")
+        resp = client.patch(
+            "/api/presence/status",
+            json={"status_message": "x" * 140},
+            headers=user["headers"],
+        )
+        assert resp.status_code == 200, resp.text
+
+    def test_null_status_message_clears_it(self):
+        client = TestClient(app)
+        user = _register(client, "presence_status_clear")
+        client.patch("/api/presence/status", json={"status_message": "hi"}, headers=user["headers"])
+        resp = client.patch("/api/presence/status", json={"status_message": None}, headers=user["headers"])
+        assert resp.status_code == 200
+        assert resp.json()["status_message"] is None
+
+
+class TestPresenceStoreTTL:
+    """Exercises the real Redis TTL semantics directly against the service,
+    with a short heartbeat_ttl so the test doesn't need to sleep for the
+    real 60s default used in production. This is the cheap, real proxy for
+    "TTL expiry means offline" the task asked for -- same code path
+    (real Redis SET ... EX / EXISTS / MGET), just a much shorter window.
+    """
+
+    async def test_ttl_expiry_flips_online_to_offline(self):
+        store = PresenceStore(heartbeat_ttl=1)
+        try:
+            uid = 987654321  # doesn't need to exist in Postgres; Redis-only check
+            assert await store.is_online(uid) is False
+
+            await store.heartbeat(uid)
+            assert await store.is_online(uid) is True
+
+            await asyncio.sleep(1.5)
+            assert await store.is_online(uid) is False
+        finally:
+            await store.close()
+
+    async def test_is_online_batch_matches_is_online_semantics(self):
+        """Regression coverage for 47ab492: is_online_batch must return a
+        set of ids usable with `in`, not a bool."""
+        store = PresenceStore(heartbeat_ttl=30)
+        try:
+            online_uid, offline_uid = 111222333, 444555666
+            await store.heartbeat(online_uid)
+
+            result = await store.is_online_batch([online_uid, offline_uid])
+            assert isinstance(result, set)
+            assert online_uid in result
+            assert offline_uid not in result
+        finally:
+            await store.close()
+
+    async def test_is_online_batch_empty_list(self):
+        store = PresenceStore()
+        try:
+            assert await store.is_online_batch([]) == set()
+        finally:
+            await store.close()