diff --git a/app.py b/app.py
index 1ec8c77..b1c221a 100644
--- a/app.py
+++ b/app.py
@@ -10,6 +10,7 @@ import time
 from pathlib import Path
 from werkzeug.utils import secure_filename
 from flask import Flask, render_template, request, redirect, url_for, flash, send_from_directory, abort, session, jsonify
+from markupsafe import escape, Markup
 from flask_login import LoginManager, login_user, logout_user, login_required, current_user
 from dotenv import load_dotenv
 from functools import lru_cache
@@ -81,15 +82,36 @@ login_manager.login_message = 'Please log in to access this page.'
 # Initialize user service
 user_service = UserService()
 
-# Initialize polling service
+# Background services (polling scheduler + filter engine) are initialized
+# lazily on the first request, NOT at import time. This keeps importing this
+# module side-effect-free (no scheduler threads, no filter-engine init), so
+# `from app import app` is safe for migration scripts and tests. The full
+# app-factory split is deferred to Phase 1.
 from polling_service import polling_service
-polling_service.init_app(app)
-polling_service.start()
-
-# Initialize filter engine
 from filter_pipeline import FilterEngine
-filter_engine = FilterEngine.get_instance()
-logger.info(f"FilterEngine initialized with {len(filter_engine.get_available_filtersets())} filtersets")
+
+filter_engine = None  # set lazily in _ensure_services_started()
+_services_started = False
+
+
+@app.before_request
+def _ensure_services_started():
+    """Start the polling scheduler and initialize the filter engine on the
+    first request, once. Kept out of module import so importing this module
+    has no side effects.
+    """
+    global filter_engine, _services_started
+    if _services_started:
+        return
+    if filter_engine is None:
+        filter_engine = FilterEngine.get_instance()
+        logger.info(
+            f"FilterEngine initialized with "
+            f"{len(filter_engine.get_available_filtersets())} filtersets"
+        )
+    polling_service.init_app(app)
+    polling_service.start()
+    _services_started = True
 
 # Initialize OAuth for Auth0
 oauth = OAuth(app)
@@ -114,6 +136,10 @@ def _is_safe_filterset(filterset):
     """Validate filterset name for security"""
     if not filterset or not isinstance(filterset, str):
         return False
+    # filter_engine is initialized lazily on the first request; if it has not
+    # been initialized yet, fail closed.
+    if filter_engine is None:
+        return False
     # Check against available filtersets from filter_engine
     allowed_filtersets = set(filter_engine.get_available_filtersets())
     return filterset in allowed_filtersets and re.match(r'^[a-zA-Z0-9_-]+$', filterset)
@@ -239,10 +265,40 @@ def _validate_user_settings(settings_str):
 # Add custom Jinja filters
 @app.template_filter('nl2br')
 def nl2br_filter(text):
-    """Convert newlines to <br> tags"""
+    """Convert newlines to <br> tags.
+
+    Escapes the input first (so raw HTML in user content cannot inject
+    markup), then inserts <br> tags and marks the result safe. Use this
+    instead of `| safe | nl2br`, which left an XSS hole.
+    """
     if not text:
         return text
-    return text.replace('\n', '<br>\n')
+    return Markup(str(escape(text)).replace('\n', '<br>\n'))
+
+
+@app.template_filter('timeago')
+def timeago_filter(timestamp):
+    """Format a unix timestamp as a relative time string ('3m ago', '5h ago',
+    '2d ago'), falling back to a date for older posts and 'Recently' for
+    missing/invalid input. Replaces the undefined `moment(...).fromNow()`
+    pattern that always rendered 'Recently'.
+    """
+    if not timestamp:
+        return 'Recently'
+    try:
+        ts = float(timestamp)
+    except (TypeError, ValueError):
+        return 'Recently'
+    diff = time.time() - ts
+    if diff < 0:
+        return 'Recently'
+    if diff < 3600:
+        return f'{int(diff // 60)}m ago'
+    if diff < 86400:
+        return f'{int(diff // 3600)}h ago'
+    if diff < 604800:
+        return f'{int(diff // 86400)}d ago'
+    return time.strftime('%Y-%m-%d', time.localtime(ts))
 
 
 @login_manager.user_loader
@@ -1062,8 +1118,8 @@ def password_reset(token):
         password = request.form.get('password', '')
         confirm_password = request.form.get('confirm_password', '')
 
-        if not password or len(password) < 6:
-            flash('Password must be at least 6 characters', 'error')
+        if not password or len(password) < MIN_PASSWORD_LENGTH:
+            flash(f'Password must be at least {MIN_PASSWORD_LENGTH} characters', 'error')
             return render_template('password_reset.html')
 
         if password != confirm_password:
diff --git a/filter_pipeline/engine.py b/filter_pipeline/engine.py
index e79e5a4..d6220e9 100644
--- a/filter_pipeline/engine.py
+++ b/filter_pipeline/engine.py
@@ -184,10 +184,16 @@ class FilterEngine:
         if self.config.is_ai_enabled():
             self._init_stages()
 
-        # If AI is disabled but filterset requires it, fall back to no_filter
+        # If AI is disabled but the filterset requires it, do NOT silently pass
+        # everything as no_filter. Pass the posts through (so the feed is not
+        # blanked) but mark every result as FAILED with an explicit error so the
+        # degradation is observable, not silent.
         if not self.config.is_ai_enabled() and filterset_name != 'no_filter':
-            logger.warning(f"AI disabled but '{filterset_name}' requires AI - falling back to 'no_filter'")
-            return self._process_no_filter(posts)
+            logger.warning(
+                f"AI disabled but filterset '{filterset_name}' requires AI - "
+                f"passing posts through unfiltered with FAILED status"
+            )
+            return self._process_ai_disabled(filterset_name, posts)
 
         # Get pipeline stages for this filterset
         stage_names = self._get_stages_for_filterset(filterset_name)
@@ -218,6 +224,28 @@ class FilterEngine:
 
         return results
 
+    def _process_ai_disabled(self, filterset_name: str, posts: List[Dict[str, Any]]) -> List[FilterResult]:
+        """Pass posts through unfiltered when the requested filterset needs AI
+        but AI is disabled. Unlike no_filter, every result is marked FAILED with
+        an explicit error so the degradation is observable rather than silent.
+        """
+        results = []
+        for post in posts:
+            result = FilterResult(
+                post_uuid=post.get('uuid', ''),
+                passed=True,  # do not blank the feed
+                score=0.5,  # neutral score
+                categories=[],
+                tags=[],
+                filterset_name=filterset_name,
+                processed_at=datetime.now(),
+                status=ProcessingStatus.FAILED,
+                error=f"AI disabled: filterset '{filterset_name}' requires AI; passed through unfiltered"
+            )
+            results.append(result)
+
+        return results
+
     def _get_stages_for_filterset(self, filterset_name: str) -> List[str]:
         """Get pipeline stages to run for a filterset"""
         filterset = self.config.get_filterset(filterset_name)
diff --git a/migrate_bookmarks.py b/migrate_bookmarks.py
index bcb6352..4c5ac6e 100644
--- a/migrate_bookmarks.py
+++ b/migrate_bookmarks.py
@@ -5,7 +5,7 @@ Migration script to create the bookmarks table.
 
 import os
 import sys
-from database import init_db
+from database import init_db, db
 from flask import Flask
 
 # Add the current directory to Python path
@@ -24,12 +24,13 @@ def main():
     app = create_app()
     
     with app.app_context():
-        # Initialize database
-        db = init_db(app)
-        
+        # Initialize database (init_db binds the shared `db` extension; it
+        # returns None, so use the module-level `db` directly).
+        init_db(app)
+
         # Import models to register them
         from models import User, Session, PollSource, PollLog, Bookmark
-        
+
         # Create all tables (will only create missing ones)
         db.create_all()
         
diff --git a/templates/post_detail.html b/templates/post_detail.html
index 29d70b8..41ab284 100644
--- a/templates/post_detail.html
+++ b/templates/post_detail.html
@@ -31,7 +31,7 @@
                     <span class="post-source">{{ post.source_display if post.source_display else ('r/' + post.source if post.platform == 'reddit' else post.source) }}</span>
                     <span class="post-separator">•</span>
                 {% endif %}
-                <span class="post-time">{{ moment(post.timestamp).fromNow() if moment else 'Recently' }}</span>
+                <span class="post-time" data-timestamp="{{ post.timestamp }}">{{ post.timestamp|timeago }}</span>
                 {% if post.url and not post.url.startswith('/') %}
                     <span class="external-link-indicator">🔗</span>
                 {% endif %}
@@ -48,7 +48,7 @@
         
         {% if post.content %}
             <div class="post-content">
-                {{ post.content | safe | nl2br }}
+                {{ post.content | nl2br }}
             </div>
         {% endif %}
         
@@ -101,10 +101,10 @@
                 <div class="comment-header">
                     <span class="comment-author">{{ comment.author }}</span>
                     <span class="comment-separator">•</span>
-                    <span class="comment-time">{{ moment(comment.timestamp).fromNow() if moment else 'Recently' }}</span>
+                    <span class="comment-time" data-timestamp="{{ comment.timestamp }}">{{ comment.timestamp|timeago }}</span>
                 </div>
                 <div class="comment-content">
-                    {{ comment.content | safe | nl2br }}
+                    {{ comment.content | nl2br }}
                 </div>
                 <div class="comment-footer">
                     <div class="comment-score">