diff --git a/docs/podcasts.md b/docs/podcasts.md
index a41b54e..10cb7b3 100644
--- a/docs/podcasts.md
+++ b/docs/podcasts.md
@@ -59,8 +59,38 @@ first), ＋ (Merkliste), ★. Filters stay: Favoriten, Verlauf, Angefangen, Geh
 
 Tabs: **Wiedergabeliste · Merkliste · Strom · Abos**.
 
-Browsing within the subscriptions (themes, people, similar episodes, search) — later, to be
-thought through.
+## Favourite subscriptions
+
+★ on a subscription (`PodcastFeed.favorite`): a mark and a filter (Abos: only the favourites;
+episode lists: `filter=fave_feeds`), whatever its level — it puts nothing on the Merkliste by
+itself. Started with the Stamm subscriptions marked.
+
+## Stöbern (decided 2026-10-19)
+
+Within the subscriptions:
+- **Search** in titles and shownotes: `GET /podcasts/discover/search/?q=…&stream=1` (all
+  words, the last as a prefix; titles weigh more). SQLite FTS5 index, contentless
+  (`podcasts/discover.py`), filled after every refresh; `manage.py index_episodes` fills it.
+- **Similar episodes** from the other podcasts: `GET /podcasts/discover/similar/<id>/`.
+- **People** (who talks, who gives a talk), from episode titles only (`podcasts/people.py`):
+  `GET /podcasts/discover/people/` → `[{name, episodes, feeds, score}]`, the ones in the most
+  listened podcasts first; `GET /podcasts/discover/person/?name=` → their episodes.
+- **Topics** of the Strom's last days, chips as for the news: `GET /podcasts/discover/topics/?days=7`
+  → `[{label, ids}]`.
+- **Forgotten**: `GET /podcasts/discover/forgotten/` → `{loved: [...], rare: [...]}` — unheard
+  episodes older than 60 days of the podcasts listened to most, and of the rare ones; the same
+  all day.
+
+One controlled step out (Apple's episode search and fyyd; no key):
+- **People elsewhere**: `GET /podcasts/discover/elsewhere/?name=` → episodes naming the person
+  in podcasts the user doesn't have (`{source, podcast, feed_url, title, audio_url, guid,
+  pub_date, duration_seconds, artwork_url, link}`). `POST /podcasts/discover/take/` with such an
+  item puts the episode on the Merkliste; its podcast becomes an *Auf Abruf* subscription.
+- **Recommendations**: `GET /podcasts/discover/recommend/` → podcasts like the most listened ones
+  (fyyd), recommended for several first: `{title, feed_url, artwork_url, description, link,
+  because: [titles], count}`. Subscribing: the usual feeds/add/ (then *Auf Abruf* or Strom).
+
+Show search stays on Apple (best coverage).
 
 ## Moving over
 
diff --git a/podcasts/discover.py b/podcasts/discover.py
new file mode 100644
index 0000000..e526f23
--- /dev/null
+++ b/podcasts/discover.py
@@ -0,0 +1,529 @@
+"""Stöbern (docs/podcasts.md): finding episodes in the user's own podcasts — search in titles
+and shownotes, similar episodes, people, topics, forgotten ones — and one controlled step out:
+where people from them are guests elsewhere (Apple's episode search, fyyd), and podcasts like
+the most listened ones (fyyd's recommendations). An episode from outside goes onto the
+Merkliste; its podcast becomes an "Auf Abruf" subscription.
+
+The full-text index is SQLite FTS5, contentless (only the index, no second copy of 100 MB of
+shownotes): podcasts_episode_fts, rowid = the episode's id, filled by index_new() after every
+refresh (new episodes have higher ids). Episodes deleted since stay in the index; every query
+joins the episodes, so they never show.
+"""
+import html
+import logging
+import math
+import random
+import re
+import unicodedata
+import urllib.parse
+from collections import Counter, defaultdict
+from concurrent.futures import ThreadPoolExecutor
+from datetime import timedelta
+
+import requests
+from django.core.cache import cache
+from django.db import connection
+from django.db.models import Q
+from django.http import JsonResponse
+from django.utils import timezone
+from django.utils.dateparse import parse_datetime
+from django.views.decorators.csrf import csrf_exempt
+from django.views.decorators.http import require_http_methods
+
+from . import people as people_mod
+from .models import EpisodeProgress, PodcastEpisode, PodcastFeed, PodcastSettings
+
+log = logging.getLogger(__name__)
+
+FTS = 'podcasts_episode_fts'
+VOCAB = 'podcasts_episode_fts_vocab'
+BODY_CHARS = 6000            # of the shownotes, per episode
+INDEX_BATCH = 2000
+SIMILAR_TERMS = 10
+HTTP_TIMEOUT = 6
+USER_AGENT = 'diora podcast discovery'
+
+CREATE_SQL = [
+    f"CREATE VIRTUAL TABLE IF NOT EXISTS {FTS} USING fts5(title, body, content='', contentless_delete=1, "
+    f"tokenize='unicode61 remove_diacritics 2')",
+    f"CREATE VIRTUAL TABLE IF NOT EXISTS {VOCAB} USING fts5vocab({FTS}, 'row')",
+]
+DROP_SQL = [f'DROP TABLE IF EXISTS {VOCAB}', f'DROP TABLE IF EXISTS {FTS}']
+
+
+def _plain(text):
+    text = re.sub(r'<[^>]+>', ' ', text or '')
+    return re.sub(r'\s+', ' ', html.unescape(text)).strip()
+
+
+def _fold(word):
+    """A word as the index keeps it: lower case, without diacritics."""
+    w = unicodedata.normalize('NFKD', word.lower())
+    return ''.join(c for c in w if not unicodedata.combining(c))
+
+
+_WORD = re.compile(r'\w+', re.UNICODE)
+
+
+def _words(text):
+    return [_fold(w) for w in _WORD.findall(text or '') if not w.isdigit()]
+
+
+def _quoted(term):
+    return '"' + term.replace('"', '""') + '"'
+
+
+# ---- the index ------------------------------------------------------------------------------------
+
+def index_new(limit=None):
+    """Index the episodes newer than the newest one indexed. Returns how many."""
+    with connection.cursor() as cur:
+        cur.execute(f'SELECT max(rowid) FROM {FTS}')
+        last = cur.fetchone()[0] or 0
+    done = 0
+    while True:
+        rows = list(PodcastEpisode.objects.filter(id__gt=last).order_by('id')
+                    .values_list('id', 'title', 'description')[:INDEX_BATCH])
+        if not rows:
+            return done
+        with connection.cursor() as cur:
+            cur.executemany(f'INSERT OR REPLACE INTO {FTS}(rowid, title, body) VALUES (%s, %s, %s)',
+                            [(i, t or '', _plain(d)[:BODY_CHARS]) for i, t, d in rows])
+        done += len(rows)
+        last = rows[-1][0]
+        if limit and done >= limit:
+            return done
+
+
+def _user_episode_sql(user, stream=False, other_than_feed=None):
+    """SQL and params restricting fts rowids to the user's podcast episodes."""
+    sql = (f' JOIN podcasts_podcastepisode e ON e.id = {FTS}.rowid'
+           f' JOIN podcasts_podcastfeed pf ON pf.id = e.feed_id'
+           f' WHERE pf.user_id = %s AND pf.kind = %s')
+    params = [user.id, PodcastFeed.PODCAST]
+    if stream:
+        sql += ' AND pf.new_episodes != %s'
+        params.append(PodcastFeed.NEW_IGNORE)
+    if other_than_feed is not None:
+        sql += ' AND e.feed_id != %s'
+        params.append(other_than_feed)
+    return sql, params
+
+
+def search_ids(user, q, stream=False, limit=50, offset=0):
+    """Episode ids of the user's podcasts matching all words of q (the last one as a prefix,
+    while typing), titles weighing three times the shownotes, best first."""
+    words = _words(q)[:12]
+    if not words:
+        return []
+    match = ' '.join(_quoted(w) for w in words[:-1]) + (' ' if len(words) > 1 else '') + _quoted(words[-1]) + '*'
+    where, params = _user_episode_sql(user, stream)
+    with connection.cursor() as cur:
+        cur.execute(f'SELECT {FTS}.rowid FROM {FTS}{where} AND {FTS} MATCH %s '
+                    f'ORDER BY bm25({FTS}, 3.0, 1.0) LIMIT %s OFFSET %s', params + [match, limit, offset])
+        return [r[0] for r in cur.fetchall()]
+
+
+def similar_ids(user, episode, limit=20):
+    """Episodes of the user's other podcasts most like `episode`: its rarest telling words
+    (tf·idf over the index) asked for together, best match first."""
+    tf = Counter()
+    for w in _words(episode.title):
+        tf[w] += 3
+    for w in _words(_plain(episode.description)[:BODY_CHARS]):
+        tf[w] += 1
+    terms = [w for w in tf if len(w) > 2]
+    if not terms:
+        return []
+    with connection.cursor() as cur:
+        cur.execute(f'SELECT count(*) FROM podcasts_podcastepisode')
+        n = cur.fetchone()[0] or 1
+        df = {}
+        for i in range(0, len(terms), 500):
+            chunk = terms[i:i + 500]
+            cur.execute(f'SELECT term, doc FROM {VOCAB} WHERE term IN ({",".join(["%s"] * len(chunk))})', chunk)
+            df.update(dict(cur.fetchall()))
+    scored = []
+    for w, c in tf.items():
+        d = df.get(w, 0)
+        if d < 2 or d > max(n * 0.05, 10):     # only here, or everywhere: says nothing
+            continue
+        scored.append((c * math.log(n / d), w))
+    scored.sort(reverse=True)
+    best = [w for _, w in scored[:SIMILAR_TERMS]]
+    if not best:
+        return []
+    where, params = _user_episode_sql(user, other_than_feed=episode.feed_id)
+    with connection.cursor() as cur:
+        cur.execute(f'SELECT {FTS}.rowid FROM {FTS}{where} AND {FTS} MATCH %s '
+                    f'ORDER BY bm25({FTS}, 3.0, 1.0) LIMIT %s',
+                    params + [' OR '.join(_quoted(w) for w in best), limit])
+        return [r[0] for r in cur.fetchall()]
+
+
+# ---- helpers ---------------------------------------------------------------------------------------
+
+def _rows(user, ids):
+    """The episodes as the episode lists give them, in the order of ids."""
+    from .views import _episode_list_data
+    if not ids:
+        return []
+    eps, _ = _episode_list_data(user, ids=ids, limit=len(ids))
+    by_id = {e['id']: e for e in eps}
+    return [by_id[i] for i in ids if i in by_id]
+
+
+def _heard_by_feed(user):
+    """{feed_id: episodes heard in the last 12 months} (as heard_365 in state/)."""
+    from .views import HEARD_MIN_SECONDS
+    since = timezone.now() - timedelta(days=365)
+    out = Counter()
+    for fid in (EpisodeProgress.objects.filter(user=user, updated_at__gte=since)
+                .filter(Q(played=True) | Q(position_seconds__gt=HEARD_MIN_SECONDS))
+                .values_list('episode__feed_id', flat=True)):
+        out[fid] += 1
+    return out
+
+
+def _norm_url(url):
+    u = (url or '').strip().lower()
+    u = re.sub(r'^https?://', '', u)
+    return u.rstrip('/')
+
+
+def _get_json(url, **params):
+    try:
+        r = requests.get(url, params=params, timeout=HTTP_TIMEOUT, headers={'User-Agent': USER_AGENT})
+        r.raise_for_status()
+        return r.json()
+    except Exception as e:   # noqa: BLE001 - a service away is no error for the user
+        log.info('discover: %s: %s', url, e)
+        return None
+
+
+# ---- people ----------------------------------------------------------------------------------------
+
+PEOPLE_TTL = 12 * 3600
+
+
+def people_of(user):
+    key = f'podcasts:people:{user.id}'
+    got = cache.get(key)
+    if got is None:
+        rows = list(PodcastEpisode.objects.filter(feed__user=user, feed__kind=PodcastFeed.PODCAST)
+                    .values_list('id', 'feed_id', 'title'))
+        got = people_mod.people(rows, _heard_by_feed(user))[:200]
+        cache.set(key, got, PEOPLE_TTL)
+    return got
+
+
+def person_ids(user, name, limit=200):
+    return list(PodcastEpisode.objects.filter(feed__user=user, feed__kind=PodcastFeed.PODCAST, title__icontains=name)
+                .order_by('-pub_date', '-id').values_list('id', flat=True)[:limit])
+
+
+def elsewhere(user, name):
+    """Episodes with `name` in podcasts the user doesn't have: Apple's episode search and fyyd's,
+    those that name the person in their title. [{source, podcast, feed_url, title, audio_url,
+    guid, pub_date, duration_seconds, artwork_url, link}]"""
+    key = f'podcasts:elsewhere:{user.id}:' + re.sub(r'\W+', '_', _fold(name))
+    got = cache.get(key)
+    if got is not None:
+        return got
+    mine = {_norm_url(u) for u in PodcastFeed.objects.filter(user=user).values_list('rss_url', flat=True)}
+    mine_titles = {t.casefold() for t in PodcastFeed.objects.filter(user=user).values_list('title', flat=True)}
+    folded = _fold(name)
+    out, seen = [], set()
+
+    def take(item):
+        if folded not in _fold(item['title']) or not item['feed_url'] or not item['audio_url']:
+            return
+        if _norm_url(item['feed_url']) in mine or item['podcast'].casefold() in mine_titles:
+            return
+        k = (_fold(item['title']), _fold(item['podcast']))
+        if k in seen:
+            return
+        seen.add(k)
+        out.append(item)
+
+    apple = _get_json('https://itunes.apple.com/search', term=name, media='podcast', entity='podcastEpisode',
+                      limit=50, country='DE') or {}
+    for r in apple.get('results', []):
+        take({'source': 'apple', 'podcast': r.get('collectionName', ''), 'feed_url': r.get('feedUrl', ''),
+              'title': r.get('trackName', ''), 'audio_url': r.get('episodeUrl', ''), 'guid': r.get('episodeGuid', ''),
+              'pub_date': r.get('releaseDate'), 'duration_seconds': int((r.get('trackTimeMillis') or 0) / 1000),
+              'artwork_url': r.get('artworkUrl600') or r.get('artworkUrl160') or '',
+              'link': r.get('trackViewUrl', '')})
+    fy = _get_json('https://api.fyyd.de/0.2/search/episode', term=f'"{name}"', count=50) or {}
+    eps = [e for e in (fy.get('data') or []) if isinstance(e, dict) and folded in _fold(e.get('title', ''))]
+    feeds = {}
+    for pid in list(dict.fromkeys(e.get('podcast_id') for e in eps))[:12]:
+        p = _get_json('https://api.fyyd.de/0.2/podcast', podcast_id=pid) or {}
+        if isinstance(p.get('data'), dict):
+            feeds[pid] = p['data']
+    for e in eps:
+        p = feeds.get(e.get('podcast_id'))
+        if not p:
+            continue
+        take({'source': 'fyyd', 'podcast': p.get('title', ''), 'feed_url': p.get('xmlURL', ''),
+              'title': html.unescape(e.get('title', '')), 'audio_url': e.get('enclosure', ''), 'guid': e.get('guid', ''),
+              'pub_date': e.get('pubdate'), 'duration_seconds': int(e.get('duration') or 0),
+              'artwork_url': e.get('imgURL') or p.get('imgURL') or '', 'link': e.get('url_fyyd', '')})
+    out.sort(key=lambda i: i['pub_date'] or '', reverse=True)
+    cache.set(key, out, 6 * 3600)
+    return out
+
+
+# ---- topics, forgotten, recommendations ------------------------------------------------------------
+
+def topics_of(user, days=7):
+    """Topics of the Strom's last days (feeds.textproc.topics, as for the news):
+    [{'label', 'ids'}], the largest first."""
+    from feeds import textproc
+    since = timezone.now() - timedelta(days=days)
+    eps = list(PodcastEpisode.objects.filter(feed__user=user, feed__kind=PodcastFeed.PODCAST, pub_date__gte=since)
+               .exclude(feed__new_episodes=PodcastFeed.NEW_IGNORE)
+               .values_list('id', 'title', 'description', 'feed__title'))
+    stories = [(i, f'{t}. {_plain(d)[:400]}', {ft}) for i, t, d, ft in eps]
+    return textproc.topics(stories, max_topics=10)
+
+
+FORGOTTEN_DAYS = 60
+
+
+def forgotten(user):
+    """Unheard older episodes of the podcasts the user listens to most, and of the rare ones —
+    a handful each, the same all day."""
+    from .views import _intervals
+    rnd = random.Random(f'{user.id}:{timezone.localdate()}')
+    heard = _heard_by_feed(user)
+    started = set(EpisodeProgress.objects.filter(user=user).filter(Q(played=True) | Q(position_seconds__gt=0))
+                  .values_list('episode_id', flat=True))
+    before = timezone.now() - timedelta(days=FORGOTTEN_DAYS)
+
+    def pick(feed_ids, n):
+        ids = [i for i in PodcastEpisode.objects.filter(feed_id__in=feed_ids, pub_date__lt=before)
+               .values_list('id', flat=True) if i not in started]
+        return rnd.sample(ids, min(n, len(ids)))
+
+    mine = list(PodcastFeed.objects.filter(user=user, kind=PodcastFeed.PODCAST)
+                .exclude(new_episodes=PodcastFeed.NEW_IGNORE).values_list('id', flat=True))
+    loved = [f for f in mine if heard.get(f, 0) >= 3]
+    intervals = _intervals(user)
+    rare = [f for f in mine if (intervals.get(f) or 0) >= 30]
+    return {'loved': _rows(user, pick(loved, 15)), 'rare': _rows(user, pick(rare, 10))}
+
+
+RECOMMEND_FROM = 10      # the most listened podcasts asked about
+RECOMMEND_TTL = 24 * 3600
+
+
+def _fyyd_id(feed):
+    got = _get_json('https://api.fyyd.de/0.2/search/podcast', title=feed['title'], count=5) or {}
+    want = _norm_url(feed['rss_url'])
+    for p in got.get('data') or []:
+        if isinstance(p, dict) and (_norm_url(p.get('xmlURL')) == want or (p.get('title') or '').casefold() == feed['title'].casefold()):
+            return p.get('id')
+    return None
+
+
+def recommend(user):
+    """Podcasts like the user's most listened (fyyd's recommendations), the ones recommended for
+    several of them first; none the user has. [{title, feed_url, artwork_url, description, link,
+    because: [titles], count}]"""
+    key = f'podcasts:recommend:{user.id}'
+    got = cache.get(key)
+    if got is not None:
+        return got
+    heard = _heard_by_feed(user)
+    feeds = list(PodcastFeed.objects.filter(user=user, kind=PodcastFeed.PODCAST).values('id', 'title', 'rss_url'))
+    top = sorted((f for f in feeds if heard.get(f['id'])), key=lambda f: -heard[f['id']])[:RECOMMEND_FROM]
+    mine = {_norm_url(f['rss_url']) for f in feeds}
+    mine_titles = {f['title'].casefold() for f in feeds}
+
+    def ask(feed):
+        pid = _fyyd_id(feed)
+        if not pid:
+            return feed, []
+        r = _get_json('https://api.fyyd.de/0.2/podcast/recommend', podcast_id=pid, count=12) or {}
+        return feed, [p for p in r.get('data') or [] if isinstance(p, dict)]
+
+    found = {}
+    with ThreadPoolExecutor(max_workers=5) as pool:
+        for feed, recs in pool.map(ask, top):
+            for p in recs:
+                url = p.get('xmlURL') or ''
+                if not url or _norm_url(url) in mine or (p.get('title') or '').casefold() in mine_titles:
+                    continue
+                k = _norm_url(url)
+                item = found.setdefault(k, {'title': p.get('title', ''), 'feed_url': url, 'artwork_url': p.get('imgURL') or '',
+                                            'description': _plain(p.get('description'))[:300],
+                                            'link': p.get('htmlURL') or '', 'because': [], 'count': 0})
+                if feed['title'] not in item['because']:
+                    item['because'].append(feed['title'])
+                    item['count'] += 1
+    out = sorted(found.values(), key=lambda i: (-i['count'], i['title'].casefold()))[:30]
+    cache.set(key, out, RECOMMEND_TTL)
+    return out
+
+
+# ---- an episode from outside -------------------------------------------------------------------------
+
+def take(user, item):
+    """An episode from elsewhere onto the Merkliste: its podcast becomes an "Auf Abruf"
+    subscription (or is one already), the episode is found in it — or, when the feed no longer
+    lists it, made from what the search said. Returns (episode, created_feed)."""
+    from .views import _later_add, _refresh_feed
+    url = (item.get('feed_url') or '').strip()
+    feed = PodcastFeed.objects.filter(user=user, rss_url=url).first()
+    created = False
+    if feed is None:
+        feed = PodcastFeed.objects.create(
+            user=user, rss_url=url, title=(item.get('podcast') or url)[:300], artwork_url=(item.get('artwork_url') or '')[:1000],
+            **{**PodcastSettings.new_feed_defaults(user), 'new_episodes': PodcastFeed.NEW_IGNORE})
+        created = True
+        try:
+            _refresh_feed(feed)
+        except Exception as e:   # noqa: BLE001 - the episode is made from the search below
+            log.info('discover: %s: %s', url, e)
+    eps = feed.episodes.all()
+    ep = None
+    if item.get('guid'):
+        ep = eps.filter(guid=item['guid']).first()
+    if ep is None and item.get('audio_url'):
+        ep = eps.filter(audio_url=item['audio_url']).first()
+    if ep is None and item.get('title'):
+        ep = eps.filter(title__iexact=item['title']).first()
+    if ep is None:
+        when = parse_datetime(item.get('pub_date') or '') if item.get('pub_date') else None
+        ep = PodcastEpisode.objects.create(
+            feed=feed, guid=(item.get('guid') or item.get('audio_url') or item.get('title'))[:1000],
+            title=(item.get('title') or '')[:500], audio_url=item.get('audio_url', '')[:1000],
+            duration_seconds=int(item.get('duration_seconds') or 0), pub_date=when,
+            artwork_url=(item.get('artwork_url') or '')[:1000])
+    _later_add(user, [ep.id])
+    try:
+        index_new()
+    except Exception:   # noqa: BLE001
+        pass
+    return ep, created
+
+
+# ---- views ---------------------------------------------------------------------------------------------
+
+def _auth(request):
+    if not request.user.is_authenticated:
+        return JsonResponse({'error': 'authentication required'}, status=401)
+    return None
+
+
+def _int(request, name, default, low, high):
+    try:
+        return min(max(int(request.GET.get(name, default)), low), high)
+    except ValueError:
+        return default
+
+
+@require_http_methods(['GET'])
+def search_view(request):
+    """?q=…&stream=1&offset=&limit=: episodes whose title or shownotes have every word."""
+    err = _auth(request)
+    if err:
+        return err
+    ids = search_ids(request.user, request.GET.get('q', ''), stream=request.GET.get('stream') == '1',
+                     limit=_int(request, 'limit', 50, 1, 200), offset=_int(request, 'offset', 0, 0, 10 ** 6))
+    return JsonResponse({'episodes': _rows(request.user, ids)})
+
+
+@require_http_methods(['GET'])
+def similar_view(request, pk):
+    err = _auth(request)
+    if err:
+        return err
+    ep = PodcastEpisode.objects.filter(pk=pk, feed__user=request.user).first()
+    if ep is None:
+        return JsonResponse({'error': 'not found'}, status=404)
+    return JsonResponse({'episodes': _rows(request.user, similar_ids(request.user, ep, _int(request, 'limit', 20, 1, 50)))})
+
+
+@require_http_methods(['GET'])
+def people_view(request):
+    """The people in the user's podcasts: [{name, episodes, feeds, score}], the most relevant first."""
+    err = _auth(request)
+    if err:
+        return err
+    return JsonResponse({'people': people_of(request.user)[:_int(request, 'limit', 80, 1, 200)]})
+
+
+@require_http_methods(['GET'])
+def person_view(request):
+    """?name=: the person's episodes in the user's podcasts, newest first."""
+    err = _auth(request)
+    if err:
+        return err
+    name = request.GET.get('name', '').strip()
+    if len(name) < 3:
+        return JsonResponse({'error': 'name required'}, status=400)
+    return JsonResponse({'name': name, 'episodes': _rows(request.user, person_ids(request.user, name))})
+
+
+@require_http_methods(['GET'])
+def elsewhere_view(request):
+    """?name=: episodes with the person in podcasts the user doesn't have."""
+    err = _auth(request)
+    if err:
+        return err
+    name = request.GET.get('name', '').strip()
+    if len(name) < 3:
+        return JsonResponse({'error': 'name required'}, status=400)
+    return JsonResponse({'name': name, 'episodes': elsewhere(request.user, name)})
+
+
+@require_http_methods(['GET'])
+def topics_view(request):
+    """?days=7: topics of the Strom's last days: [{label, ids}], plus their episodes."""
+    err = _auth(request)
+    if err:
+        return err
+    days = _int(request, 'days', 7, 1, 30)
+    key = f'podcasts:topics:{request.user.id}:{days}'
+    got = cache.get(key)
+    if got is None:
+        got = topics_of(request.user, days)
+        cache.set(key, got, 3600)
+    return JsonResponse({'topics': got})
+
+
+@require_http_methods(['GET'])
+def forgotten_view(request):
+    err = _auth(request)
+    if err:
+        return err
+    return JsonResponse(forgotten(request.user))
+
+
+@require_http_methods(['GET'])
+def recommend_view(request):
+    err = _auth(request)
+    if err:
+        return err
+    return JsonResponse({'podcasts': recommend(request.user)})
+
+
+@csrf_exempt
+@require_http_methods(['POST'])
+def take_view(request):
+    """{podcast, feed_url, title, audio_url, guid, pub_date, duration_seconds, artwork_url} (an
+    item of elsewhere/): onto the Merkliste; returns {episode_id, feed_id, subscribed}."""
+    import json
+    err = _auth(request)
+    if err:
+        return err
+    try:
+        item = json.loads(request.body or b'{}')
+    except (json.JSONDecodeError, ValueError):
+        return JsonResponse({'error': 'invalid JSON'}, status=400)
+    if not isinstance(item, dict) or not re.match(r'^https?://', item.get('feed_url') or '') or not item.get('audio_url'):
+        return JsonResponse({'error': 'feed_url and audio_url required'}, status=400)
+    ep, created = take(request.user, item)
+    return JsonResponse({'ok': True, 'episode_id': ep.id, 'feed_id': ep.feed_id, 'subscribed': created})
diff --git a/podcasts/management/commands/index_episodes.py b/podcasts/management/commands/index_episodes.py
new file mode 100644
index 0000000..228dc7a
--- /dev/null
+++ b/podcasts/management/commands/index_episodes.py
@@ -0,0 +1,10 @@
+from django.core.management.base import BaseCommand
+
+from podcasts.discover import index_new
+
+
+class Command(BaseCommand):
+    help = 'Put the episodes not yet in the full-text index into it (podcasts/discover.py)'
+
+    def handle(self, *args, **options):
+        self.stdout.write(f'{index_new()} episodes indexed')
diff --git a/podcasts/management/commands/refresh_feeds.py b/podcasts/management/commands/refresh_feeds.py
index 8049cad..4ec9dfd 100644
--- a/podcasts/management/commands/refresh_feeds.py
+++ b/podcasts/management/commands/refresh_feeds.py
@@ -2,6 +2,7 @@ from django.core.management.base import BaseCommand
 from django.utils import timezone
 
 from podcasts.models import PodcastFeed
+from podcasts.discover import index_new
 from podcasts.views import _in_backoff_q, _refresh_feeds, refresh_due
 
 
@@ -40,4 +41,6 @@ class Command(BaseCommand):
 
         # Parallel downloads, one per distinct feed URL (see _refresh_feeds).
         refreshed, failed, new = _refresh_feeds(qs, full=options['full'])
-        self.stdout.write(f'{refreshed} feeds refreshed, {failed} failed, +{new} episodes')
+        # New episodes into the full-text index (Stöbern, podcasts/discover.py).
+        indexed = index_new()
+        self.stdout.write(f'{refreshed} feeds refreshed, {failed} failed, +{new} episodes, {indexed} indexed')
diff --git a/podcasts/migrations/0023_episode_fts.py b/podcasts/migrations/0023_episode_fts.py
new file mode 100644
index 0000000..41ad84d
--- /dev/null
+++ b/podcasts/migrations/0023_episode_fts.py
@@ -0,0 +1,27 @@
+# The full-text index for Stöbern (podcasts/discover.py): SQLite FTS5, contentless.
+from django.db import migrations
+
+from podcasts.discover import CREATE_SQL, DROP_SQL
+
+
+def create(apps, schema_editor):
+    if schema_editor.connection.vendor != 'sqlite':
+        return
+    for sql in CREATE_SQL:
+        schema_editor.execute(sql)
+
+
+def drop(apps, schema_editor):
+    if schema_editor.connection.vendor != 'sqlite':
+        return
+    for sql in DROP_SQL:
+        schema_editor.execute(sql)
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('podcasts', '0022_lists_and_levels'),
+    ]
+
+    operations = [migrations.RunPython(create, drop)]
diff --git a/podcasts/migrations/0024_feed_favorite.py b/podcasts/migrations/0024_feed_favorite.py
new file mode 100644
index 0000000..f882e87
--- /dev/null
+++ b/podcasts/migrations/0024_feed_favorite.py
@@ -0,0 +1,18 @@
+# Generated by Django 6.1 on 2026-10-19 16:45
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('podcasts', '0023_episode_fts'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='podcastfeed',
+            name='favorite',
+            field=models.BooleanField(default=False),
+        ),
+    ]
diff --git a/podcasts/models.py b/podcasts/models.py
index d71ed96..ddb7700 100644
--- a/podcasts/models.py
+++ b/podcasts/models.py
@@ -67,6 +67,9 @@ class PodcastFeed(models.Model):
     # New episodes: what happens to them, and which ones count at all.
     new_episodes = models.CharField(max_length=10, choices=NEW_EPISODE_ACTIONS, default=NEW_INBOX)
     refresh_interval = models.CharField(max_length=10, choices=REFRESH_INTERVALS, default='', blank=True)
+    # A favourite subscription (★ in the Abos): only a mark and a filter, whatever its level —
+    # it puts nothing on the Merkliste by itself (docs/podcasts.md).
+    favorite = models.BooleanField(default=False)
     # Comma-separated words, case-insensitive, matched against the title.
     include_words = models.CharField(max_length=500, blank=True)
     exclude_words = models.CharField(max_length=500, blank=True)
diff --git a/podcasts/people.py b/podcasts/people.py
new file mode 100644
index 0000000..a7f0096
--- /dev/null
+++ b/podcasts/people.py
@@ -0,0 +1,122 @@
+"""People in the user's podcasts (docs/podcasts.md, Stöbern): who is talked with, who gives a
+talk — found in episode titles and descriptions, no service asked.
+
+A name is two or three capitalised words ("Peter Decker", "Ulrike Eifler", "Jean-Luc Mélenchon"),
+possibly with a particle ("Wolfgang von Goethe"). Two kinds of evidence:
+- strong: right after a cue that announces a person — "mit", "w/", "with", "feat.", "Gast:",
+  "im Gespräch mit", "interview with", "von" in a talk's title, " - Name - " between dashes;
+- weak: anywhere else in a title or description.
+A name counts when it has strong evidence in at least STRONG_MIN episodes, or turns up in at least
+FEEDS_MIN podcasts (a person known across the user's world). Words that are no names (places,
+institutions, weekdays, sentence starts) are left out by a list and by the rule that a name's
+first word must not be one of them.
+"""
+import html
+import re
+from collections import Counter, defaultdict
+
+_W = r"[A-ZÄÖÜÀ-Ý][a-zäöüßà-ÿ]+(?:-[A-ZÄÖÜÀ-Ý][a-zäöüßà-ÿ]+)?"
+_PART = r"(?:von|van|der|de|da|di|du|la|le|zu|ten|del|bin|al)"
+_NAME = rf"{_W}(?:\s+{_PART})?(?:\s+{_W}){{1,2}}"
+NAME_RE = re.compile(rf"(?<![\w\-])({_NAME})(?![\w])")
+CUE_RE = re.compile(
+    r"(?i:\bmit\b|\bw/|\bwith\b|\bfeat\.?|\bft\.|\bGast:|\bGäste:|\bguest:|\binterview(?:t)? with\b|"
+    r"\bim Gespräch mit\b|\bin conversation with\b|\bvon\b|\bby\b)\s*(" + _NAME + r")|\s[-–|]\s(" + _NAME + r")")
+# (only the cue ignores case: a name's words must start with a capital)
+
+# First or last words that make a capitalised pair no person.
+NOT_NAME_WORDS = set('''
+Der Die Das Den Dem Des Ein Eine Einen Einem Einer Und Oder Aber Mit Von Nach Bei Für Auf Aus Im In Am Zum Zur
+Vor Seit Als Wie Wenn Teil Folge Episode Staffel Kapitel Band Live Special Bonus Trailer Podcast Interview
+Gespräch Vortrag Diskussion Lesung Hörspiel Feature Interviews Folgen Talk Show News Update Week Day Days
+The A An And Or But With From For After Before On At By To Of As It This That These Those Our Your Their
+Mr Mrs Ms Dr Prof Herr Frau Doktor Professor Sir Saint Sankt St
+Monday Tuesday Wednesday Thursday Friday Saturday Sunday Montag Dienstag Mittwoch Donnerstag Freitag
+Samstag Sonntag January February March April May June July August September October November December
+Januar Februar März Juni Juli Oktober Dezember Frühling Sommer Herbst Winter
+Deutschland Deutschlands Deutsche Deutschen Deutscher Europa Europas Europäische Europäischen Amerika
+Amerikas Usa China Chinas Russland Russlands Ukraine Israel Gaza Iran Frankreich Großbritannien England
+Berlin Hamburg München Wien Zürich Köln Leipzig Dresden London Paris New York Washington Moskau
+United States Kingdom Union Republic Republik Bank Partei Linke Linken Grüne Grünen Spd Cdu Afd Fdp Bsw
+Universität University College Institute Institut Stiftung Verlag Radio Deutschlandfunk Kultur Nova
+Zeitung Magazin Review Times Post News Press Media Netflix Youtube Spotify Apple Google Amazon Twitter
+Google Meta Tesla Marxismus Kapitalismus Sozialismus Kommunismus Faschismus Imperialismus Krieg Frieden
+Gott Jesus Christus Weltkrieg Revolution Kalter Kalten Krise Wahl Wahlen Bundestag Bundesregierung
+Remaster Remastered Audio Enhanced Edition Version Full Part Chapter Book Season Classic Klassiker
+Hegel Marx Engels Kant Nietzsche Freud Lacan
+Doku Hd Reupload Reaction Re View Neue Neuer Zu Gast Gäste My Last Free Speech Open Source Big Tech Cup World
+War Files Kanzler Bundeskanzler Präsident Präsidentin President Minister Ministerin Trumps Künstliche
+White House Middle East Silicon Valley Wall Street Mental Health Red Thread Critical Role Trio Infernal
+Asian Japan Nhk Arte Spiegel Zeit Faz Taz Bild Planet Money Lage Nation Inside Austria Today Daily Weekly
+Political Science Letzte Foreign Private Golden Keine Tag El Los Las San Santa Rio Fort Port Lake Mount
+'''.split())
+STRONG_MIN = 2
+FEEDS_MIN = 3
+
+
+def plain(text):
+    text = re.sub(r'<[^>]+>', ' ', text or '')
+    return re.sub(r'\s+', ' ', html.unescape(text)).strip()
+
+
+def _clean(name):
+    words = name.split()
+    # "Mit Peter Decker" etc.: drop leading words that are no name.
+    while words and words[0] in NOT_NAME_WORDS:
+        words = words[1:]
+    while words and words[-1] in NOT_NAME_WORDS:
+        words = words[:-1]
+    caps = [w for w in words if w[0].isupper()]
+    if len(caps) < 2 or any(w in NOT_NAME_WORDS for w in caps):
+        return None
+    if any(len(w) > 1 and w.isupper() for w in caps):   # "WAS TUN", "ARTE Reportage": no person
+        return None
+    return ' '.join(words)
+
+
+def mentions(title):
+    """(strong, weak): the names in one episode's title."""
+    strong, weak = set(), set()
+    for m in CUE_RE.finditer(title or ''):
+        n = _clean(m.group(1) or m.group(2))
+        if n:
+            strong.add(n)
+    for m in NAME_RE.finditer(title or ''):
+        n = _clean(m.group(1))
+        if n:
+            weak.add(n)
+    return strong, weak - strong
+
+
+def people(episodes, weights=None):
+    """episodes: [(episode_id, feed_id, title)]; weights: {feed_id: how much the user listens}
+    (heard_365). The people found, the most relevant first: [{'name', 'episodes', 'feeds'}].
+
+    Only titles: descriptions carry credits and boilerplate. First names are learnt from the
+    strong mentions ("mit Peter Decker": Peter); a name found elsewhere in a title counts only
+    with such a first name."""
+    found = [(ep_id, feed_id, *mentions(title)) for ep_id, feed_id, title in episodes]
+    surnames = defaultdict(set)
+    for _, _, strong, _ in found:
+        for n in strong:
+            w = n.split()
+            surnames[w[0]].add(w[-1])
+    # A first name: one that comes with at least two different surnames in strong mentions.
+    first_names = {f for f, last in surnames.items() if len(last) >= 2 and f not in NOT_NAME_WORDS}
+    eps = defaultdict(set)
+    feeds = defaultdict(set)
+    strong_n = Counter()
+    for ep_id, feed_id, strong, weak in found:
+        for n in strong:
+            strong_n[n] += 1
+        for n in {n for n in strong | weak if n.split()[0] in first_names}:
+            eps[n].add(ep_id)
+            feeds[n].add(feed_id)
+    weights = weights or {}
+    out = []
+    for n, ids in eps.items():
+        if strong_n[n] >= STRONG_MIN or len(feeds[n]) >= FEEDS_MIN:
+            score = len(ids) + 5 * sum(min(weights.get(f, 0), 50) for f in feeds[n]) ** 0.5
+            out.append({'name': n, 'episodes': len(ids), 'feeds': len(feeds[n]), 'score': round(score, 1)})
+    out.sort(key=lambda p: (-p['score'], p['name']))
+    return out
diff --git a/podcasts/test_discover.py b/podcasts/test_discover.py
new file mode 100644
index 0000000..0419eaa
--- /dev/null
+++ b/podcasts/test_discover.py
@@ -0,0 +1,149 @@
+"""Stöbern (podcasts/discover.py): the full-text index, similar episodes, people, an episode
+from outside onto the Merkliste."""
+import json
+from datetime import datetime, timedelta, timezone
+from unittest.mock import patch
+
+from django.contrib.auth.models import User
+from django.core.cache import cache
+from django.test import TestCase
+
+from . import discover, people
+from .models import EpisodeProgress, PodcastEpisode, PodcastFeed, PodcastLater
+
+
+class PeopleTests(TestCase):
+    def test_names_from_titles(self):
+        eps = [
+            (1, 1, 'Episode 651: Die Brandmauer mit Peter Decker'),
+            (2, 1, 'Episode 568: Trumps Zollkrieg mit Peter Decker'),
+            (3, 2, 'Imperialismus heute - Peter Decker - 2006'),
+            (4, 3, 'Gespräch mit Peter Schadt über Arbeit'),
+            (5, 3, 'Ulrike Eifler im Gespräch mit Peter Thiel'),
+            (6, 4, 'Was tun? Mit Ulrike Herrmann'),
+            (9, 4, 'Streik! Diskussion mit Ulrike Eifler'),
+            (7, 5, 'Gespräch mit Ulrike Herrmann über das Ende des Kapitalismus'),
+            (8, 6, 'Live aus Berlin: Die Linke und Friedrich Merz'),
+        ]
+        found = {p['name']: p for p in people.people(eps)}
+        self.assertEqual(found['Peter Decker']['episodes'], 3)
+        self.assertEqual(found['Peter Decker']['feeds'], 2)
+        self.assertIn('Ulrike Herrmann', found)
+        self.assertNotIn('Die Brandmauer', found)
+        self.assertNotIn('Die Linke', found)
+
+
+class IndexTests(TestCase):
+    def setUp(self):
+        cache.clear()
+        self.user = User.objects.create_user(username='finder', password='pw12345678')
+        self.client.force_login(self.user)
+        self.other = User.objects.create_user(username='other', password='pw12345678')
+        self.f1 = PodcastFeed.objects.create(user=self.user, rss_url='https://example.org/1', title='Eins')
+        self.f2 = PodcastFeed.objects.create(user=self.user, rss_url='https://example.org/2', title='Zwei')
+        self.f3 = PodcastFeed.objects.create(user=self.user, rss_url='https://example.org/3', title='Drei',
+                                             new_episodes=PodcastFeed.NEW_IGNORE)
+        theirs = PodcastFeed.objects.create(user=self.other, rss_url='https://example.org/1', title='Eins')
+
+        def ep(feed, guid, title, desc=''):
+            return PodcastEpisode.objects.create(feed=feed, guid=guid, title=title, description=desc,
+                                                 audio_url=f'https://example.org/{guid}.mp3',
+                                                 pub_date=datetime(2026, 1, 1, tzinfo=timezone.utc))
+        self.a = ep(self.f1, 'a', 'Über die Ökonomie der Plattformen', '<p>Mit Gästen über <b>Plattformkapitalismus</b> und Arbeit.</p>')
+        self.b = ep(self.f2, 'b', 'Plattformkapitalismus erklärt', 'Wie Plattformen Arbeit organisieren.')
+        self.c = ep(self.f2, 'c', 'Gartenarbeit im Herbst', 'Tomaten, Kürbisse und Laub.')
+        self.d = ep(self.f3, 'd', 'Noch mehr Plattformkapitalismus', 'Auf Abruf.')
+        for i in range(30):   # words everywhere say nothing
+            ep(self.f1, f'x{i}', f'Folge {i}', 'Arbeit und Gesellschaft')
+        ep(theirs, 't', 'Plattformkapitalismus bei jemand anderem', '')
+        self.assertGreaterEqual(discover.index_new(), 35)
+        self.assertEqual(discover.index_new(), 0, 'nothing twice')
+
+    def test_search_titles_shownotes_prefix_folding(self):
+        r = self.client.get('/podcasts/discover/search/?q=plattformkap').json()
+        self.assertEqual({e['id'] for e in r['episodes']}, {self.a.id, self.b.id, self.d.id})
+        self.assertEqual(r['episodes'][0]['id'] in (self.b.id, self.d.id), True, 'a title match first')
+        r = self.client.get('/podcasts/discover/search/?q=okonomie').json()
+        self.assertEqual([e['id'] for e in r['episodes']], [self.a.id])
+        r = self.client.get('/podcasts/discover/search/?q=plattformkapitalismus&stream=1').json()
+        self.assertNotIn(self.d.id, [e['id'] for e in r['episodes']], 'Auf Abruf is not in the Strom')
+
+    def test_similar_from_other_podcasts(self):
+        r = self.client.get(f'/podcasts/discover/similar/{self.b.id}/').json()
+        ids = [e['id'] for e in r['episodes']]
+        self.assertIn(self.a.id, ids)
+        self.assertNotIn(self.c.id, ids, 'same podcast, and about gardens')
+        self.assertEqual(self.client.get('/podcasts/discover/similar/999999/').status_code, 404)
+
+    def test_person_episodes(self):
+        PodcastEpisode.objects.create(feed=self.f1, guid='p', title='Arbeit mit Peter Decker', audio_url='https://example.org/p.mp3')
+        r = self.client.get('/podcasts/discover/person/?name=Peter Decker').json()
+        self.assertEqual([e['title'] for e in r['episodes']], ['Arbeit mit Peter Decker'])
+
+    def test_forgotten(self):
+        old = PodcastEpisode.objects.create(feed=self.f2, guid='old', title='Alt', audio_url='https://example.org/old.mp3',
+                                            pub_date=datetime.now(timezone.utc) - timedelta(days=200))
+        for e in (self.b, self.c, self.a):
+            EpisodeProgress.objects.create(user=self.user, episode=e, played=True)
+        PodcastEpisode.objects.create(feed=self.f2, guid='h1', title='H1', audio_url='https://example.org/h1.mp3')
+        heard = PodcastEpisode.objects.create(feed=self.f2, guid='h2', title='H2', audio_url='https://example.org/h2.mp3')
+        EpisodeProgress.objects.create(user=self.user, episode=heard, played=True)
+        r = self.client.get('/podcasts/discover/forgotten/').json()
+        self.assertEqual([e['id'] for e in r['loved']], [old.id])
+
+
+class OutsideTests(TestCase):
+    def setUp(self):
+        cache.clear()
+        self.user = User.objects.create_user(username='out', password='pw12345678')
+        self.client.force_login(self.user)
+        PodcastFeed.objects.create(user=self.user, rss_url='https://mine.example/feed', title='Mein Podcast')
+
+    def test_elsewhere_keeps_only_the_person_and_unknown_podcasts(self):
+        apple = {'results': [
+            {'collectionName': 'Fremd', 'feedUrl': 'https://other.example/feed', 'trackName': 'Gespräch mit Peter Decker',
+             'episodeUrl': 'https://other.example/1.mp3', 'episodeGuid': 'g1', 'releaseDate': '2026-01-02T00:00:00Z',
+             'trackTimeMillis': 60000},
+            {'collectionName': 'Mein Podcast', 'feedUrl': 'https://mine.example/feed', 'trackName': 'Mit Peter Decker',
+             'episodeUrl': 'https://mine.example/1.mp3'},
+            {'collectionName': 'Fremd', 'feedUrl': 'https://other.example/feed', 'trackName': 'Peter Pan',
+             'episodeUrl': 'https://other.example/2.mp3'},
+        ]}
+
+        def fake(url, **params):
+            return apple if 'itunes' in url else {'data': []}
+        with patch.object(discover, '_get_json', fake):
+            r = self.client.get('/podcasts/discover/elsewhere/?name=Peter Decker').json()
+        self.assertEqual([e['title'] for e in r['episodes']], ['Gespräch mit Peter Decker'])
+        self.assertEqual(r['episodes'][0]['duration_seconds'], 60)
+
+    def test_take_subscribes_auf_abruf_and_merklists(self):
+        item = {'podcast': 'Fremd', 'feed_url': 'https://other.example/feed', 'title': 'Gespräch mit Peter Decker',
+                'audio_url': 'https://other.example/1.mp3', 'guid': 'g1', 'pub_date': '2026-01-02T00:00:00Z',
+                'duration_seconds': 60}
+        with patch('podcasts.views._refresh_feed', side_effect=OSError('offline')):
+            r = self.client.post('/podcasts/discover/take/', json.dumps(item), content_type='application/json').json()
+        self.assertTrue(r['subscribed'])
+        feed = PodcastFeed.objects.get(pk=r['feed_id'])
+        self.assertEqual(feed.new_episodes, PodcastFeed.NEW_IGNORE)
+        self.assertTrue(PodcastLater.objects.filter(user=self.user, episode_id=r['episode_id']).exists())
+        with patch('podcasts.views._refresh_feed', side_effect=OSError('offline')):
+            r2 = self.client.post('/podcasts/discover/take/', json.dumps(item), content_type='application/json').json()
+        self.assertEqual((r2['episode_id'], r2['subscribed']), (r['episode_id'], False))
+
+    def test_recommend_counts_and_leaves_out_mine(self):
+        EpisodeProgress.objects.create(
+            user=self.user, played=True,
+            episode=PodcastEpisode.objects.create(feed=PodcastFeed.objects.get(title='Mein Podcast'), guid='m',
+                                                  title='M', audio_url='https://mine.example/m.mp3'))
+
+        def fake(url, **params):
+            if url.endswith('search/podcast'):
+                return {'data': [{'id': 7, 'title': 'Mein Podcast', 'xmlURL': 'https://mine.example/feed'}]}
+            if url.endswith('recommend'):
+                return {'data': [{'id': 8, 'title': 'Neu', 'xmlURL': 'https://new.example/feed'},
+                                 {'id': 7, 'title': 'Mein Podcast', 'xmlURL': 'https://mine.example/feed'}]}
+            return None
+        with patch.object(discover, '_get_json', fake):
+            r = self.client.get('/podcasts/discover/recommend/').json()
+        self.assertEqual([(p['title'], p['because']) for p in r['podcasts']], [('Neu', ['Mein Podcast'])])
diff --git a/podcasts/test_lists.py b/podcasts/test_lists.py
index 947f744..6fe4c19 100644
--- a/podcasts/test_lists.py
+++ b/podcasts/test_lists.py
@@ -151,6 +151,16 @@ class IntervalTests(Base):
         due = refresh_due(list(PodcastFeed.objects.filter(pk__in=[daily.pk, stale.pk, new.pk])), now)
         self.assertEqual(set(due), {stale.pk, new.pk})
 
+    def test_favourite_subscriptions(self):
+        r = self.post(f'/podcasts/feeds/{self.feed.pk}/settings/', {'favorite': True}).json()
+        self.assertTrue(r['favorite'])
+        other = PodcastFeed.objects.create(user=self.user, rss_url='https://example.org/o', title='O')
+        PodcastEpisode.objects.create(feed=other, guid='o', title='O1', audio_url='https://example.org/o.mp3')
+        got = self.client.get('/podcasts/episodes/?filter=fave_feeds&limit=50').json()
+        self.assertEqual({e['feed__id'] for e in got['episodes']}, {self.feed.pk})
+        self.assertTrue([f for f in self.state()['feeds'] if f['id'] == self.feed.pk][0]['favorite'])
+        self.assertEqual(PodcastLater.objects.filter(user=self.user).count(), 0, 'a favourite queues nothing')
+
     def test_setting_it(self):
         r = self.post(f'/podcasts/feeds/{self.feed.pk}/settings/', {'refresh_interval': 'weekly'}).json()
         self.assertEqual(r['refresh_interval'], 'weekly')
diff --git a/podcasts/tests.py b/podcasts/tests.py
index 702b16d..4265b61 100644
--- a/podcasts/tests.py
+++ b/podcasts/tests.py
@@ -1106,7 +1106,7 @@ FEED_DEFAULTS = {
     'playback_speed': None, 'skip_intro': 0, 'skip_outro': 0, 'volume': 1.0, 'new_episodes': 'inbox',
     'include_words': '', 'exclude_words': '', 'min_duration': 0, 'auto_queue': False, 'auto_download': False,
     'notify': False, 'episode_sort': '', 'episode_filter': [], 'skip_silence': None, 'kind': 'podcast',
-    'folder': '', 'folders': [], 'refresh_interval': '',
+    'folder': '', 'folders': [], 'refresh_interval': '', 'favorite': False,
 }
 
 
@@ -1134,7 +1134,7 @@ class FeedSettingsTests(TestCase):
             'new_episodes': 'inbox', 'include_words': 'News, Talk', 'exclude_words': 'Teaser', 'min_duration': 10,
             'auto_queue': False, 'auto_download': False, 'notify': False,
             'episode_sort': '', 'episode_filter': [], 'skip_silence': None, 'kind': 'podcast', 'folder': '', 'folders': [],
-            'refresh_interval': '',
+            'refresh_interval': '', 'favorite': False,
         })
         self.feed.refresh_from_db()
         self.assertEqual((self.feed.playback_speed, self.feed.skip_intro, self.feed.volume, self.feed.min_duration),
diff --git a/podcasts/urls.py b/podcasts/urls.py
index f548b79..2433919 100644
--- a/podcasts/urls.py
+++ b/podcasts/urls.py
@@ -1,5 +1,5 @@
 from django.urls import path
-from . import audiobooks, folders, views
+from . import audiobooks, discover, folders, views
 
 urlpatterns = [
     path('search/', views.podcast_search, name='podcast_search'),
@@ -39,6 +39,16 @@ urlpatterns = [
     path('later/remove/', views.later_remove, name='podcast_later_remove'),
     path('lists/', views.lists_switch, name='podcast_lists_switch'),
     path('stream/seen/', views.stream_seen, name='podcast_stream_seen'),
+    # Stöbern (docs/podcasts.md, podcasts/discover.py)
+    path('discover/search/', discover.search_view, name='podcast_discover_search'),
+    path('discover/similar/<int:pk>/', discover.similar_view, name='podcast_discover_similar'),
+    path('discover/people/', discover.people_view, name='podcast_discover_people'),
+    path('discover/person/', discover.person_view, name='podcast_discover_person'),
+    path('discover/elsewhere/', discover.elsewhere_view, name='podcast_discover_elsewhere'),
+    path('discover/topics/', discover.topics_view, name='podcast_discover_topics'),
+    path('discover/forgotten/', discover.forgotten_view, name='podcast_discover_forgotten'),
+    path('discover/recommend/', discover.recommend_view, name='podcast_discover_recommend'),
+    path('discover/take/', discover.take_view, name='podcast_discover_take'),
     path('settings/', views.podcast_settings, name='podcast_settings'),
     path('progress/save/', views.save_progress, name='podcast_save_progress'),
     path('progress/mark-played/', views.mark_played, name='podcast_mark_played'),
diff --git a/podcasts/views.py b/podcasts/views.py
index 6c712fb..3d2a9e7 100644
--- a/podcasts/views.py
+++ b/podcasts/views.py
@@ -892,7 +892,7 @@ FEED_SETTING_FIELDS = (
     'playback_speed', 'skip_intro', 'skip_outro', 'volume',
     'new_episodes', 'include_words', 'exclude_words', 'min_duration', 'auto_queue', 'auto_download',
     'notify', 'episode_sort', 'episode_filter', 'skip_silence', 'kind', 'folder', 'folders',
-    'refresh_interval',
+    'refresh_interval', 'favorite',
 )
 
 
@@ -1186,6 +1186,7 @@ FEED_SETTING_PARSERS = {
     'folders': folders_mod.clean_list,
     'skip_silence': _optional_boolean,
     'refresh_interval': lambda v: _refresh_interval(v),
+    'favorite': _boolean,
 }
 
 
@@ -1599,6 +1600,8 @@ EPISODE_FILTERS = {
     'later': Q(in_later=True),
     # The Strom (docs/podcasts.md): every subscription but the "Auf Abruf" ones.
     'stream': ~Q(feed__new_episodes=PodcastFeed.NEW_IGNORE),
+    # Episodes of the favourite subscriptions.
+    'fave_feeds': Q(feed__favorite=True),
 }
 # Episodes of unknown length (0) come last when sorting by length.
 _UNKNOWN_LENGTH_LAST = Case(When(duration_seconds__gt=0, then=Value(0)), default=Value(1), output_field=IntegerField())
diff --git a/static/js/sw.js b/static/js/sw.js
index 461a9dd..42ee1f0 100644
--- a/static/js/sw.js
+++ b/static/js/sw.js
@@ -2,7 +2,7 @@
  * diora service worker — caches the app shell for offline use.
  */
 
-const CACHE = 'diora-v137';
+const CACHE = 'diora-v138';
 const SHELL = [
   '/static/css/app.css',
   '/static/js/app/01-core.js',