diff --git a/crudkit/dsl.py b/crudkit/dsl.py
index 5a4e5b8..a9ee931 100644
--- a/crudkit/dsl.py
+++ b/crudkit/dsl.py
@@ -108,11 +108,6 @@ def build_query(Model, spec: QuerySpec, eager_policy=None):
         col = getattr(Model, key[1:] if desc_ else key)
         stmt = stmt.order_by(desc(col) if desc_ else asc(col))
 
-    if not spec.order_by and spec.page and spec.per_page:
-        pk_cols = inspect(Model).primary_key
-        if pk_cols:
-            stmt = stmt.order_by(*(asc(c) for c in pk_cols))
-
     # eager loading
     if eager_policy:
         opts = eager_policy(Model, spec.expand)
diff --git a/crudkit/html/templates/crudkit/_macros.html b/crudkit/html/templates/crudkit/_macros.html
index 1bcdfd4..f713e81 100644
--- a/crudkit/html/templates/crudkit/_macros.html
+++ b/crudkit/html/templates/crudkit/_macros.html
@@ -31,15 +31,17 @@
 {%- endfor -%}
 {%- endmacro %}
 
-{% macro pager(model, page, pages, per_page, sort, filters, fields_csv) -%}
-<nav id="pager" hx-swap-oob="true">
-    <a hx-get="/ui/{{ model }}/frag/rows?page={{ page-1 }}&per_page={{ per_page }}{% if sort %}&sort={{ sort }}{% endif %}{% if fields_csv %}&fields_csv={{ fields_csv|urlencode }}{% endif %}{% for k,v in filters.items() %}&{{k}}={{ v|urlencode }}{% endfor %}"
-        hx-target="#rows" rel="prev" hx-swap="innerHTML">Prev</a>
-
+{% macro pager(model, page, pages, per_page, sort, filters) -%}
+<nav class="pager">
+    {%- if page > 1 -%}
+    <a hx-get="/{{ model }}/frag/rows?page={{ page-1 }}&per_page={{ per_page }}{% if sort %}&sort={{ sort }}{% endif %}{% for k,v in filters.items() %}&{{k}}={{v}}{% endfor %}"
+        hx-target="#rows" hx-push-url="true">Prev</a>
+    {%- endif -%}
     <span>Page {{ page }} / {{ pages }}</span>
-
-    <a hx-get="/ui/{{ model }}/frag/rows?page={{ page+1 }}&per_page={{ per_page }}{% if sort %}&sort={{ sort }}{% endif %}{% if fields_csv %}&fields_csv={{ fields_csv|urlencode }}{% endif %}{% for k,v in filters.items() %}&{{k}}={{ v|urlencode }}{% endfor %}"
-        hx-target="#rows" rel="next" hx-swap="innerHTML">Next</a>
+    {%- if page < pages -%} <a
+        hx-get="/{{ model }}/frag/rows?page={{ page+1 }}&per_page={{ per_page }}{% if sort %}&sort={{ sort }}{% endif %}{% for k,v in filters.items() %}&{{k}}={{v}}{% endfor %}"
+        hx-target="#rows" hx-push-url="true">Next</a>
+        {%- endif -%}
 </nav>
 {%- endmacro %}
 
diff --git a/crudkit/html/templates/crudkit/form.html b/crudkit/html/templates/crudkit/form.html
index 761658a..5f3bfb3 100644
--- a/crudkit/html/templates/crudkit/form.html
+++ b/crudkit/html/templates/crudkit/form.html
@@ -1,3 +1,3 @@
-{% import "crudkit/_macros.html" as ui %}
+{% import "_macros.html" as ui %}
 {% set action = url_for('frags.save', model=model) %}
 {{ ui.form(schema, action, method="POST", obj_id=obj.id if obj else None, hx=true) }}
\ No newline at end of file
diff --git a/crudkit/html/templates/crudkit/lis.html b/crudkit/html/templates/crudkit/lis.html
index 3ce62e7..e9b1813 100644
--- a/crudkit/html/templates/crudkit/lis.html
+++ b/crudkit/html/templates/crudkit/lis.html
@@ -1,2 +1,2 @@
-{% import "crudkit/_macros.html" as ui %}
+{% import "_macros.html" as ui %}
 {{ ui.lis(items, label_path=label_path, sublabel_path=sublabel_path, getp=getp) }}
diff --git a/crudkit/html/templates/crudkit/options.html b/crudkit/html/templates/crudkit/options.html
index 0f66d2d..34d6a2b 100644
--- a/crudkit/html/templates/crudkit/options.html
+++ b/crudkit/html/templates/crudkit/options.html
@@ -1,3 +1,3 @@
 {# Renders only <option>...</option> rows #}
-{% import "crudkit/_macros.html" as ui %}
+{% import "_macros.html" as ui %}
 {{ ui.options(items, value_attr=value_attr, label_path=label_path, getp=getp) }}
diff --git a/crudkit/html/templates/crudkit/pager.html b/crudkit/html/templates/crudkit/pager.html
deleted file mode 100644
index b3e22a3..0000000
--- a/crudkit/html/templates/crudkit/pager.html
+++ /dev/null
@@ -1,2 +0,0 @@
-{% import 'crudkit/_macros.html' as ui %}
-{{ ui.pager(model, page, pages, per_page, sort, filters, fields_csv) }}
diff --git a/crudkit/html/templates/crudkit/row.html b/crudkit/html/templates/crudkit/row.html
index 719c0ba..a3ac629 100644
--- a/crudkit/html/templates/crudkit/row.html
+++ b/crudkit/html/templates/crudkit/row.html
@@ -1,2 +1,2 @@
-{% import "crudkit/_macros.html" as ui %}
+{% import "_macros.html" as ui %}
 {{ ui.rows([obj], fields, getp=getp) }}
\ No newline at end of file
diff --git a/crudkit/html/templates/crudkit/rows.html b/crudkit/html/templates/crudkit/rows.html
index 7784328..7fafa3a 100644
--- a/crudkit/html/templates/crudkit/rows.html
+++ b/crudkit/html/templates/crudkit/rows.html
@@ -1,2 +1,3 @@
-{% import "crudkit/_macros.html" as ui %}
+{% import "_macros.html" as ui %}
 {{ ui.rows(items, fields, getp=getp) }}
+{{ ui.pager(model, page, pages, per_page, sort, filters) }}
diff --git a/crudkit/html/ui_fragments.py b/crudkit/html/ui_fragments.py
index d89bb7a..525c6e5 100644
--- a/crudkit/html/ui_fragments.py
+++ b/crudkit/html/ui_fragments.py
@@ -20,7 +20,7 @@ def make_fragments_blueprint(db_session_factory, registry: Dict[str, Any], *, na
         GET /<model>/frag/rows      -> <tr>...</tr> + pager markup if wanted
         GET /<model>/frag/form      -> <form>...</form> (auto-generated)
     """
-    bp = Blueprint(name, __name__, template_folder="templates")
+    bp = Blueprint(name, __name__, template_folder="templates/crudkit")
     def session(): return scoped_session(db_session_factory)()
 
     def _parse_filters(args):
@@ -74,7 +74,7 @@ def make_fragments_blueprint(db_session_factory, registry: Dict[str, Any], *, na
         s = session(); svc = CrudService(s, default_eager_policy)
         items, _ = svc.list(Model, spec)
 
-        return render_template("crudkit/options.html", items=items, value_attr=value_attr, label_path=label_path, getp=_getp)
+        return render_template("options.html", items=items, value_attr=value_attr, label_path=label_path, getp=_getp)
 
     @bp.get("/<model>/frag/lis")
     def lis(model):
@@ -91,7 +91,7 @@ def make_fragments_blueprint(db_session_factory, registry: Dict[str, Any], *, na
         s = session(); svc = CrudService(s, default_eager_policy)
         rows, total = svc.list(Model, spec)
         pages = (ceil(total / per_page) if page and per_page else 1)
-        return render_template("crudkit/lis.html", items=rows, label_path=label_path, sublabel_path=sublabel_path, page=page or 1, per_page=per_page or 1, total=total, model=model, sort=sort, filters=filters, getp=_getp)
+        return render_template("lis.html", items=rows, label_path=label_path, sublabel_path=sublabel_path, page=page or 1, per_page=per_page or 1, total=total, model=model, sort=sort, filters=filters, getp=_getp)
 
     @bp.get("/<model>/frag/rows")
     def rows(model):
@@ -108,12 +108,7 @@ def make_fragments_blueprint(db_session_factory, registry: Dict[str, Any], *, na
         s = session(); svc = CrudService(s, default_eager_policy)
         rows, total = svc.list(Model, spec)
         pages = max(1, ceil(total / per_page))
-
-        rows_html = render_template("crudkit/rows.html", items=rows, fields=fields, getp=_getp)
-        pager_html = render_template("crudkit/pager.html", model=model, page=page, pages=pages,
-                                     per_page=per_page, sort=sort, filters=filters, fields_csv=fields_csv)
-
-        return rows_html + pager_html
+        return render_template("rows.html", items=rows, fields=fields, page=page, pages=pages, per_page=per_page, total=total, model=model, sort=sort, filters=filters, getp=_getp)
 
     @bp.get("/<model>/frag/form")
     def form(model):
@@ -128,7 +123,7 @@ def make_fragments_blueprint(db_session_factory, registry: Dict[str, Any], *, na
         schema = build_form_schema(Model, s, obj=obj, include=include)
 
         hx = request.args.get("hx", type=int) == 1
-        return render_template("crudkit/form.html", model=model, obj=obj, schema=schema, hx=hx)
+        return render_template("form.html", model=model, obj=obj, schema=schema, hx=hx)
 
     def coerce_form_types(Model, data: dict) -> dict:
         """Turn HTML string inputs into the Python types your columns expect."""
diff --git a/inventory/__init__.py b/inventory/__init__.py
index 2ff9f2d..acd5102 100644
--- a/inventory/__init__.py
+++ b/inventory/__init__.py
@@ -14,6 +14,13 @@ if not logger.handlers:
     handler.setFormatter(logging.Formatter('%(asctime)s [%(levelname)s] %(message)s'))
     logger.addHandler(handler)
 
+def is_in_memory_sqlite():
+    uri = current_app.config.get("SQLALCHEMY_DATABASE_URI")
+    if not uri:
+        return False
+    url = make_url(uri)
+    return url.get_backend_name() == "sqlite" and url.database == ":memory:"
+
 def create_app():
     from config import Config
     app = Flask(__name__)
@@ -23,7 +30,8 @@ def create_app():
 
     with app.app_context():
         from . import models
-        db.create_all()
+        if is_in_memory_sqlite():
+            db.create_all()
 
         # ✅ db.engine is only safe to touch inside an app context
         SessionLocal = sessionmaker(bind=db.engine, expire_on_commit=False)
@@ -31,11 +39,13 @@ def create_app():
         from .models import registry
         from .routes import main
         from .routes.images import image_bp
+        from .ui.blueprint import bp as ui_bp
         from crudkit.blueprint import make_blueprint as make_json_bp
         from crudkit.html import make_fragments_blueprint as make_html_bp
 
         app.register_blueprint(main)
         app.register_blueprint(image_bp)
+        app.register_blueprint(ui_bp)
         app.register_blueprint(make_json_bp(SessionLocal, registry), url_prefix="/api")
         app.register_blueprint(make_html_bp(SessionLocal, registry), url_prefix="/ui")
 
diff --git a/inventory/routes/__init__.py b/inventory/routes/__init__.py
index 558c7db..da18176 100644
--- a/inventory/routes/__init__.py
+++ b/inventory/routes/__init__.py
@@ -11,6 +11,8 @@ log = logging.getLogger(__name__)
 
 from . import inventory, user, worklog, settings, index, search, hooks
 from .. import db
+from ..ui.blueprint import get_model_class, call
+from ..ui.defaults import default_query
 
 def _eager_from_fields(Model, fields: Iterable[str]):
     rels = {f.split(".", 1)[0] for f in fields if "." in f}
diff --git a/inventory/routes/inventory.py b/inventory/routes/inventory.py
index 30e8564..40b80f5 100644
--- a/inventory/routes/inventory.py
+++ b/inventory/routes/inventory.py
@@ -44,10 +44,7 @@ def list_inventory():
     inventory = sorted(inventory, key=lambda i: i.identifier)
 
     rows=[{"id": item.id, "cells": [row_fn(item) for row_fn in inventory_headers.values()]} for item in inventory]
-    if rows:
-        fields = [d['field'] for d in rows[0]['cells']]
-    else:
-        fields = []
+    fields = [d['field'] for d in rows[0]['cells']]
 
     return render_template(
         'table.html',
diff --git a/inventory/static/js/table.js b/inventory/static/js/table.js
index 21fe1d9..d4a3380 100644
--- a/inventory/static/js/table.js
+++ b/inventory/static/js/table.js
@@ -22,12 +22,15 @@ function Table(cfg) {
       const u = new URL(this.refreshUrl, window.location.origin);
 
       // We want server-side pagination with page/per_page
+      u.searchParams.set('view', 'table');
       u.searchParams.set('page', this.page);
       u.searchParams.set('per_page', this.perPage);
 
       // Send requested fields in the way your backend expects
+      // If your route supports &field=... repeaters, do this:
+      this.fields.forEach(f => u.searchParams.append('field', f));
       // If your route only supports "fields=a,b,c", then use:
-      if (this.fields.length) u.searchParams.set('fields_csv', this.fields.join(','));
+      // if (this.fields.length) u.searchParams.set('fields', this.fields.join(','));
 
       return u.toString();
     },
diff --git a/inventory/templates/fragments/_table_fragment.html b/inventory/templates/fragments/_table_fragment.html
index f923155..ca3d8d4 100644
--- a/inventory/templates/fragments/_table_fragment.html
+++ b/inventory/templates/fragments/_table_fragment.html
@@ -51,16 +51,17 @@
 {% endif %}
 {% endmacro %}
 
-{% macro dynamic_table(id, headers=none, fields=none, entry_route=None, title=None, page=1, per_page=15, offset=0,
-refresh_url=none, model=none) %}
+{% macro dynamic_table(id, headers=none, fields=none, entry_route=None, title=None, per_page=15, offset=0,
+refresh_url=none) %}
 <!-- Table Fragment -->
 
+{% if rows or refresh_url %}
 {% if title %}
 <label for="datatable-{{ id|default('table')|replace(' ', '-')|lower }}" class="form-label">{{ title }}</label>
 {% endif %}
 <div class="table-responsive" id="table-container-{{ id }}" x-data='Table({
         id: "{{ id }}",
-        refreshUrl: null,
+        refreshUrl: {{ refresh_url|tojson if refresh_url else "null" }},
         headers: {{ headers|tojson if headers else "[]" }},
         perPage: {{ per_page }},
         offset: {{ offset if offset else 0 }},
@@ -68,6 +69,17 @@ refresh_url=none, model=none) %}
     })'>
     <table id="datatable-{{ id|default('table')|replace(' ', '-')|lower }}"
         class="table table-bordered table-sm table-hover table-striped table-light m-0 caption-bottom">
+        <caption class="p-0">
+            <nav class="d-flex flex-column align-items-center px-2 py-1">
+                <!-- This is your pagination control -->
+                <ul class="pagination mb-0" x-ref="pagination"></ul>
+                <!-- This is just Alpine text binding -->
+                <div>
+                    Page <span x-text="page"></span> of <span x-text="pages"></span>
+                    (<span x-text="total"></span> total)
+                </div>
+            </nav>
+        </caption>
         <thead class="sticky-top">
             <tr>
                 {% for h in headers %}
@@ -75,13 +87,10 @@ refresh_url=none, model=none) %}
                 {% endfor %}
             </tr>
         </thead>
-        <tbody id="rows"
-            hx-get="/ui/{{ model }}/frag/rows?page={{ page }}&{{ per_page }}&fields_csv={{ fields|join(',') }}"
-            hx-trigger="load"
-            hx-target="#rows"
-            hx-swap="innerHTML">
-        </tbody>
+        <tbody x-ref="body"></tbody>
     </table>
-    <div id="pager"></div>
 </div>
+{% else %}
+<div class="container text-center">No data.</div>
+{% endif %}
 {% endmacro %}
\ No newline at end of file
diff --git a/inventory/templates/layout.html b/inventory/templates/layout.html
index bffff99..eb3d108 100644
--- a/inventory/templates/layout.html
+++ b/inventory/templates/layout.html
@@ -36,7 +36,7 @@
     </style>
 </head>
 
-<body class="bg-tertiary text-primary-emphasis" hx-debug="true">
+<body class="bg-tertiary text-primary-emphasis">
     <nav class="navbar navbar-expand bg-body-tertiary border-bottom">
         <div class="container-fluid">
             <a class="navbar-brand" href="{{ url_for('main.index') }}">
diff --git a/inventory/templates/table.html b/inventory/templates/table.html
index 30889a8..683fcf7 100644
--- a/inventory/templates/table.html
+++ b/inventory/templates/table.html
@@ -30,8 +30,8 @@
     id='table',
     headers=header.keys()|list if header else [],
     entry_route=entry_route,
+    refresh_url = url_for('ui.list_items', model_name=model_name, view='table'),
     offset=offset,
-    fields=fields,
-    model=model_name
+    fields=fields
 ) }}
 {% endblock %}
diff --git a/inventory/ui/blueprint.py b/inventory/ui/blueprint.py
new file mode 100644
index 0000000..683ff47
--- /dev/null
+++ b/inventory/ui/blueprint.py
@@ -0,0 +1,455 @@
+from collections import defaultdict
+
+from flask import Blueprint, request, render_template, jsonify, abort, make_response
+from sqlalchemy.engine import ScalarResult
+from sqlalchemy.exc import IntegrityError
+from sqlalchemy.orm import class_mapper, load_only, selectinload, joinedload, Load
+from sqlalchemy.sql import Select
+from typing import Any, List, cast, Iterable, Tuple, Set, Dict
+
+from .defaults import (
+    default_query, default_create, default_update, default_delete, default_serialize, default_values, default_value, default_select, ensure_order_by, count_for
+)
+
+from .. import db
+
+bp = Blueprint("ui", __name__, url_prefix="/ui")
+
+from sqlalchemy.orm import Load
+
+def _option_targets_rel(opt: Load, Model, rel_name: str) -> bool:
+    """
+    Return True if this Load option targets Model.rel_name at its root path.
+    Works for joinedload/selectinload/subqueryload options.
+    """
+    try:
+        # opt.path is a PathRegistry; .path is a tuple of (mapper, prop, mapper, prop, ...)
+        path = tuple(getattr(opt, "path", ()).path)  # type: ignore[attr-defined]
+    except Exception:
+        return False
+    if not path:
+        return False
+    # We only care about the first hop: (Mapper[Model], RelationshipProperty(rel_name))
+    if len(path) < 2:
+        return False
+    first_mapper, first_prop = path[0], path[1]
+    try:
+        is_model = first_mapper.class_ is Model  # type: ignore[attr-defined]
+        is_rel = getattr(first_prop, "key", "") == rel_name
+        return bool(is_model and is_rel)
+    except Exception:
+        return False
+
+def _has_loader_for(stmt: Select, Model, rel_name: str) -> bool:
+    """
+    True if stmt already has any loader option configured for Model.rel_name.
+    """
+    opts = getattr(stmt, "_with_options", ())  # SQLAlchemy stores Load options here
+    for opt in opts:
+        if isinstance(opt, Load) and _option_targets_rel(opt, Model, rel_name):
+            return True
+    return False
+
+def _strategy_for_rel_attr(rel_attr) -> type[Load] | None:
+    # rel_attr is an InstrumentedAttribute (Model.foo)
+    prop = getattr(rel_attr, "property", None)
+    lazy = getattr(prop, "lazy", None)
+    if lazy in ("joined", "subquery"):
+        return joinedload
+    if lazy == "selectin":
+        return selectinload
+    # default if mapper left it None or something exotic like 'raise'
+    return selectinload
+
+def apply_model_default_eager(stmt: Select, Model, skip_rels: Set[str]) -> Select:
+    # mapper.relationships yields RelationshipProperty objects
+    mapper = class_mapper(Model)
+    for prop in mapper.relationships:
+        if prop.key in skip_rels:
+            continue
+        lazy = getattr(prop, "lazy", None)
+        if lazy in ("joined", "subquery"):
+            stmt = stmt.options(joinedload(getattr(Model, prop.key)))
+        elif lazy == "selectin":
+            stmt = stmt.options(selectinload(getattr(Model, prop.key)))
+        # else: leave it alone (noload/raise/dynamic/etc.)
+    return stmt
+
+def split_fields(Model, fields: Iterable[str]) -> Tuple[Set[str], Dict[str, Set[str]]]:
+    """
+    Split requested fields into base model columns and relation->attr sets.
+    Example: ["name", "brand.name", "owner.identifier"] =>
+      base_cols = {"name"}
+      rel_cols = {"brand": {"name"}, "owner": {"identifier"}}
+    """
+    base_cols: Set[str] = set()
+    rel_cols: Dict[str, Set[str]] = defaultdict(set)
+
+    for f in fields:
+        f = f.strip()
+        if not f:
+            continue
+        if "." in f:
+            rel, attr = f.split(".", 1)
+            rel_cols[rel].add(attr)
+        else:
+            base_cols.add(f)
+    return base_cols, rel_cols
+
+def _load_only_existing(Model, names: Set[str]):
+    """
+    Return a list of mapped column attributes present on Model for load_only(...).
+    Skips relationships and unmapped/hybrid attributes so SQLA doesn’t scream.
+    """
+    cols = []
+    mapper = class_mapper(Model)
+    mapped_attr_names = set(mapper.attrs.keys())
+    for n in names:
+        if n in mapped_attr_names:
+            attr = getattr(Model, n)
+            prop = getattr(attr, "property", None)
+            if prop is not None and hasattr(prop, "columns"):
+                cols.append(attr)
+    return cols
+
+def apply_field_loaders(stmt: Select, Model, fields: Iterable[str]) -> Select:
+    base_cols, rel_cols = split_fields(Model, fields)
+
+    base_only = _load_only_existing(Model, base_cols)
+    if base_only:
+        stmt = stmt.options(load_only(*base_only))
+
+    for rel_name, attrs in rel_cols.items():
+        if not hasattr(Model, rel_name):
+            continue
+
+        # If someone already attached a loader for this relation, don't add another
+        if _has_loader_for(stmt, Model, rel_name):
+            # still allow trimming columns on the related entity if we can
+            rel_attr = getattr(Model, rel_name)
+            try:
+                target_cls = rel_attr.property.mapper.class_
+            except Exception:
+                continue
+            rel_only = _load_only_existing(target_cls, attrs)
+            if rel_only:
+                # attach a Load that only applies load_only to that path,
+                # without picking a different strategy
+                # This relies on SQLA merging load_only onto existing Load for the same path.
+                stmt = stmt.options(
+                    getattr(Load(Model), rel_name).load_only(*rel_only)
+                )
+            continue
+
+        # Otherwise choose a strategy and add it
+        rel_attr = getattr(Model, rel_name)
+        strategy = _strategy_for_rel_attr(rel_attr)
+        if not strategy:
+            continue
+        opt = strategy(rel_attr)
+
+        # Trim columns on the related entity if requested
+        try:
+            target_cls = rel_attr.property.mapper.class_
+        except Exception:
+            continue
+        rel_only = _load_only_existing(target_cls, attrs)
+        if rel_only:
+            opt = opt.options(load_only(*rel_only))
+
+        stmt = stmt.options(opt)
+
+    return stmt
+
+def _normalize(s: str) -> str:
+    return s.replace("_", "").replace("-", "").lower()
+
+def get_model_class(model_name: str) -> type:
+    """Resolve a model class by name across SA/Flask-SA versions."""
+    target = _normalize(model_name)
+
+    # SA 2.x / Flask-SQLAlchemy 3.x path
+    registry = getattr(db.Model, "registry", None)
+    if registry and getattr(registry, "mappers", None):
+        for mapper in registry.mappers:
+            cls = mapper.class_
+            # match on class name w/ and w/o underscores
+            if _normalize(cls.__name__) == target or cls.__name__.lower() == model_name.lower():
+                return cls
+
+    # Legacy Flask-SQLAlchemy 2.x path (if someone runs old stack)
+    decl = getattr(db.Model, "_decl_class_registry", None)
+    if decl:
+        for cls in decl.values():
+            if isinstance(cls, type) and (
+                _normalize(cls.__name__) == target or cls.__name__.lower() == model_name.lower()
+            ):
+                return cls
+
+    abort(404, f"Unknown resource '{model_name}'")
+
+def call(Model: type, name: str, *args: Any, **kwargs: Any) -> Any:
+    fn = getattr(Model, name, None)
+    return fn(*args, **kwargs) if callable(fn) else None
+
+from flask import request, jsonify, render_template
+from sqlalchemy.sql import Select
+from sqlalchemy.engine import ScalarResult
+from typing import Any, cast
+
+@bp.get("/<model_name>/list")
+def list_items(model_name):
+    Model = get_model_class(model_name)
+
+    text = (request.args.get("q") or "").strip() or None
+    fields_raw = (request.args.get("fields") or "").strip()
+    fields = [f.strip() for f in fields_raw.split(",") if f.strip()]
+    fields.extend(request.args.getlist("field"))
+
+    # legacy params
+    limit_param = request.args.get("limit")
+    if limit_param in (None, "", "0", "-1"):
+        effective_limit = 0
+    else:
+        effective_limit = min(int(limit_param), 500)
+
+    offset = int(request.args.get("offset", 0))
+
+    # new-school params
+    page = request.args.get("page", type=int)
+    per_page = request.args.get("per_page", type=int)
+
+    # map legacy limit/offset to page/per_page if new params not provided
+    if per_page is None:
+        per_page = effective_limit or 20  # default page size if not unlimited
+    if page is None:
+        page = (offset // per_page) + 1 if per_page else 1
+
+    # unlimited: treat as "no pagination"
+    unlimited = (per_page == 0)
+
+    view = (request.args.get("view") or "json").strip()
+    sort = (request.args.get("sort") or "").strip() or None
+    direction = (request.args.get("dir") or request.args.get("direction") or "asc").lower()
+    if direction not in ("asc", "desc"):
+        direction = "asc"
+
+    qkwargs: dict[str, Any] = {
+        "text": text,
+        "limit": 0 if unlimited else per_page,
+        "offset": 0 if unlimited else (page - 1) * per_page if per_page else 0,
+        "sort": sort,
+        "direction": direction,
+    }
+
+    # compute requested relations once
+    base_cols, rel_cols = split_fields(Model, fields)
+    skip_rels = set(rel_cols.keys()) if fields else set()
+
+    # 1) per-model override first
+    rows_any: Any = call(Model, "ui_query", db.session, **qkwargs)
+
+    stmt: Select | None = None
+    total: int
+
+    if rows_any is None:
+        stmt = default_select(Model, text=text, sort=sort, direction=direction, eager=False)
+
+        if not fields:
+            stmt = apply_model_default_eager(stmt, Model, skip_rels=set())
+        else:
+            stmt = apply_field_loaders(stmt, Model, fields)
+
+        stmt = ensure_order_by(stmt, Model, sort=sort, direction=direction)
+
+    elif isinstance(rows_any, Select):
+        # TRUST ui_query; don't add loaders on top
+        stmt = ensure_order_by(rows_any, Model, sort=sort, direction=direction)
+
+    elif isinstance(rows_any, list):
+        # materialized list; paginate in python
+        total = len(rows_any)
+        if unlimited:
+            rows = rows_any
+        else:
+            start = (page - 1) * per_page
+            end = start + per_page
+            rows = rows_any[start:end]
+        # serialize and return at the bottom like usual
+    else:
+        # SQLAlchemy Result-like or generic iterable
+        scalars = getattr(rows_any, "scalars", None)
+        if callable(scalars):
+            all_rows = list(cast(ScalarResult[Any], scalars()))
+            total = len(all_rows)
+            rows = all_rows if unlimited else all_rows[(page - 1) * per_page : (page * per_page)]
+        else:
+            try:
+                all_rows = list(rows_any)
+                total = len(all_rows)
+                rows = all_rows if unlimited else all_rows[(page - 1) * per_page : (page * per_page)]
+            except TypeError:
+                total = 1
+                rows = [rows_any]
+
+    # If we have a real Select, run it once (unlimited) or paginate once.
+    if stmt is not None:
+        if unlimited:
+            rows = list(db.session.execute(stmt).scalars())
+            total = count_for(db.session, stmt)
+        else:
+            pagination = db.paginate(stmt, page=page, per_page=per_page, error_out=False)
+            rows = pagination.items
+            total = pagination.total
+
+    # Serialize
+    if fields:
+        items = []
+        for r in rows:
+            row = {"id": r.id}
+            for f in fields:
+                if '.' in f:
+                    rel, attr = f.split('.', 1)
+                    rel_obj = getattr(r, rel, None)
+                    row[f] = getattr(rel_obj, attr, None) if rel_obj else None
+                else:
+                    row[f] = getattr(r, f, None)
+            items.append(row)
+    else:
+        items = [
+            (call(Model, "ui_serialize", r, view=view) or default_serialize(Model, r, view=view))
+            for r in rows
+        ]
+
+    # Views
+    want_option = (request.args.get("view") == "option")
+    want_list = (request.args.get("view") == "list")
+    want_table = (request.args.get("view") == "table")
+
+    if want_option:
+        return render_template("fragments/_option_fragment.html", options=items)
+    if want_list:
+        return render_template("fragments/_list_fragment.html", options=items)
+    if want_table:
+        resp = make_response(render_template("fragments/_table_data_fragment.html",
+                                            rows=items, model_name=model_name))
+        resp.headers['X-Total'] = str(total)
+        resp.headers['X-Page'] = str(page)
+        resp.headers['X-PAges'] = str((0 if unlimited else ((total + per_page - 1) // per_page)))
+        resp.headers['X-Per-Page'] = str(per_page)
+        return resp
+
+    return jsonify({
+        "items": items,
+        "total": total,
+        "page": page,
+        "per_page": per_page,
+        "pages": (0 if unlimited else ((total + per_page - 1) // per_page))
+    })
+
+@bp.post("/<model_name>/create")
+def create_item(model_name):
+    Model = get_model_class(model_name)
+    payload: dict[str, Any] = request.get_json(silent=True) or {}
+    if not payload:
+        return jsonify({"error": "Payload required"}), 422
+    try:
+        obj = call(Model, 'ui_create', db.session, payload=payload) \
+            or default_create(db.session, Model, payload)
+    except IntegrityError:
+        db.session.rollback()
+        return jsonify({"error": "Duplicate"}), 409
+    data = call(Model, 'ui_serialize', obj) or default_serialize(Model, obj)
+    want_html = (request.args.get('view') == 'option') or ("HX-Request" in request.headers)
+    if want_html:
+        return "Yo."
+    return jsonify(data), 201
+
+@bp.post("/<model_name>/update")
+def update_item(model_name):
+    Model = get_model_class(model_name)
+    payload: dict[str, Any] = request.get_json(silent=True) or {}
+
+    id_raw: Any = payload.get("id")
+    if isinstance(id_raw, bool):  # bool is an int subclass; explicitly ban
+        return jsonify({"error": "Invalid id"}), 422
+    try:
+        id_ = int(id_raw)  # will raise on None, '', junk
+    except (TypeError, ValueError):
+        return jsonify({"error": "Invalid id"}), 422
+
+    obj = call(Model, 'ui_update', db.session, id_=id_, payload=payload) \
+        or default_update(db.session, Model, id_, payload)
+    if not obj:
+        return jsonify({"error": "Not found"}), 404
+    return ("", 204)
+
+@bp.post("/<model_name>/delete")
+def delete_item(model_name):
+    Model = get_model_class(model_name)
+    payload: dict[str, Any] = request.get_json(silent=True) or {}
+    ids_raw = payload.get("ids") or []
+    if not isinstance(ids_raw, list):
+        return jsonify({"error": "Invalid ids"}), 422
+    try:
+        ids: List[int] = [int(x) for x in ids_raw]
+    except (TypeError, ValueError):
+        return jsonify({"error": "Invalid ids"}), 422
+    try:
+        deleted = call(Model, 'ui_delete', db.session, ids=ids) \
+            or default_delete(db.session, Model, ids)
+    except IntegrityError as e:
+        db.session.rollback()
+        return jsonify({"error": "Constraint", "detail": str(e)}), 409
+    return jsonify({"deleted": deleted}), 200
+
+@bp.get("/<model_name>/value")
+def get_value(model_name):
+    Model = get_model_class(model_name)
+
+    field = (request.args.get("field") or "").strip()
+    if not field:
+        return jsonify({"error": "field required"}), 422
+
+    id_raw = request.args.get("id")
+    try:
+        id_ = int(id_raw)
+    except (TypeError, ValueError):
+        return jsonify({"error": "Invalid id"}), 422
+
+    # per-model override hook: ui_value(session, id_: int, field: str) -> Any
+    try:
+        val = call(Model, "ui_value", db.session, id_=id_, field=field)
+        if val is None:
+            val = default_value(db.session, Model, id_=id_, field=field)
+    except ValueError as e:
+        return jsonify({"error": str(e)}), 400
+
+    # If HTMX hit this, keep the response boring and small
+    if request.headers.get("HX-Request"):
+        # text/plain keeps htmx happy for innerHTML swaps
+        return (str(val) if val is not None else ""), 200, {"Content-Type": "text/plain; charset=utf-8"}
+
+    return jsonify({"id": id_, "field": field, "value": val})
+
+@bp.get("<model_name>/values")
+def get_values(model_name):
+    Model = get_model_class(model_name)
+
+    raw = request.args.get("fields") or ""
+    parts = [p for p in raw.split(",") if p.strip()]
+    parts.extend(request.args.getlist("field"))
+
+    id_raw = request.args.get("id")
+    try:
+        id_ = int(id_raw)
+    except (TypeError, ValueError):
+        return jsonify({"error": "Invalid id"}), 422
+
+    try:
+        data = call(Model, "ui_values", db.session, id_=id_, fields=parts) \
+            or default_values(db.session, Model, id_=id_, fields=parts)
+    except ValueError as e:
+        return jsonify({"error": str(e)}), 400
+
+    return jsonify({"id": id_, "fields": parts, "values": data})
\ No newline at end of file
diff --git a/inventory/ui/defaults.py b/inventory/ui/defaults.py
new file mode 100644
index 0000000..ab6be3e
--- /dev/null
+++ b/inventory/ui/defaults.py
@@ -0,0 +1,356 @@
+from sqlalchemy import select, asc as sa_asc, desc as sa_desc, or_, func
+from sqlalchemy.inspection import inspect
+from sqlalchemy.orm import class_mapper, joinedload, selectinload
+from sqlalchemy.sql import Select
+from sqlalchemy.sql.sqltypes import String, Unicode, Text
+from typing import Any, Optional, cast, Iterable
+
+PREFERRED_LABELS = ("identifier", "name", "first_name", "last_name", "description")
+
+def _columns_for_text_search(Model):
+    mapper = inspect(Model)
+    cols = []
+    for c in mapper.columns:
+        if isinstance(c.type, (String, Unicode, Text)):
+            cols.append(getattr(Model, c.key))
+
+    return cols
+
+def _mapped_column(Model, attr):
+    """Return the mapped column attr on the class (InstrumentedAttribute) or None"""
+    mapper = inspect(Model)
+    if attr in mapper.columns.keys():
+        return getattr(Model, attr)
+    for prop in mapper.column_attrs:
+        if prop.key == attr:
+            return getattr(Model, prop.key)
+    return None
+
+def infer_label_attr(Model):
+    explicit = getattr(Model, 'ui_label_attr', None)
+    if explicit:
+        if _mapped_column(Model, explicit) is not None:
+            return explicit
+        raise RuntimeError(f"ui_label_attr '{explicit}' on {Model.__name__} is not a mapped column")
+
+    for a in PREFERRED_LABELS:
+        if _mapped_column(Model, a) is not None:
+            return a
+    raise RuntimeError(f"No label-like mapped column on {Model.__name__} (tried {PREFERRED_LABELS})")
+
+def count_for(session, stmt: Select) -> int:
+    # strip ORDER BY for efficiency
+    subq = stmt.order_by(None).subquery()
+    count_stmt = select(func.count()).select_from(subq)
+    return session.execute(count_stmt).scalar_one()
+
+def ensure_order_by(stmt, Model, sort=None, direction="asc"):
+    try:
+        has_order = bool(getattr(stmt, '_order_by_clauses', None))
+    except Exception:
+        has_order = False
+    if has_order:
+        return stmt
+
+    cols = []
+
+    if sort and hasattr(Model, sort):
+        col = getattr(Model, sort)
+        cols.append(col.desc() if direction == "desc" else col.asc())
+
+    if not cols:
+        ui_order_cols = getattr(Model, 'ui_order_cols', ())
+        for name in ui_order_cols or ():
+            c = getattr(Model, name, None)
+            if c is not None:
+                cols.append(c.asc())
+
+    if not cols:
+        for pk_col in inspect(Model).primary_key:
+            cols.append(pk_col.asc())
+
+    return stmt.order_by(*cols)
+
+def default_select(
+    Model,
+    *,
+    text: Optional[str] = None,
+    sort: Optional[str] = None,
+    direction: str = "asc",
+    eager = False,
+    skip_rels=frozenset()
+) -> Select[Any]:
+    stmt: Select[Any] = select(Model)
+
+    # search
+    ui_search = getattr(Model, "ui_search", None)
+    if callable(ui_search) and text:
+        stmt = cast(Select[Any], ui_search(stmt, text))
+    elif text:
+        # optional generic search fallback if you used this in default_query
+        t = f"%{text}%"
+        text_cols = _columns_for_text_search(Model)  # your existing helper
+        if text_cols:
+            stmt = stmt.where(or_(*(col.ilike(t) for col in text_cols)))
+
+    # sorting
+    if sort:
+        ui_sort = getattr(Model, "ui_sort", None)
+        if callable(ui_sort):
+            stmt = cast(Select[Any], ui_sort(stmt, sort, direction))
+        else:
+            col = getattr(Model, sort, None)
+            if col is not None:
+                stmt = stmt.order_by(sa_desc(col) if direction == "desc" else sa_asc(col))
+    else:
+        ui_order_cols = getattr(Model, "ui_order_cols", ())
+        if ui_order_cols:
+            order_cols = []
+            for name in ui_order_cols:
+                col = getattr(Model, name, None)
+                if col is not None:
+                    order_cols.append(sa_asc(col))
+            if order_cols:
+                stmt = stmt.order_by(*order_cols)
+
+    # eagerload defaults
+    opts_attr = getattr(Model, "ui_eagerload", ())
+    if callable(opts_attr):
+        opts = cast(Iterable[Any], opts_attr())   # if you prefer, pass Model in
+    else:
+        opts = cast(Iterable[Any], opts_attr)
+    for opt in opts:
+        stmt = stmt.options(opt)
+
+    if eager:
+        for prop in class_mapper(Model).relationships:
+            if prop.key in skip_rels:
+                continue
+            lazy = getattr(prop, "lazy", None)
+            if lazy in ("joined", "subquery"):
+                stmt = stmt.options(joinedload(getattr(Model, prop.key)))
+            elif lazy == "selectin":
+                stmt = stmt.options(selectinload(getattr(Model, prop.key)))
+    return stmt
+
+def default_query(
+    session,
+    Model,
+    *,
+    text: Optional[str] = None,
+    limit: int = 0,
+    offset: int = 0,
+    sort: Optional[str] = None,
+    direction: str = "asc",
+) -> list[Any]:
+    """
+    SA 2.x ONLY. Returns list[Model].
+
+    Hooks:
+      - ui_search(stmt: Select, text: str) -> Select
+      - ui_sort(stmt: Select, sort: str, direction: str) -> Select
+      - ui_order_cols: tuple[str, ...]  # default ordering columns
+    """
+    stmt: Select[Any] = select(Model)
+
+    ui_search = getattr(Model, "ui_search", None)
+    if callable(ui_search) and text:
+        stmt = cast(Select[Any], ui_search(stmt, text))
+    elif text:
+        t = f"%{text}%"
+        text_cols = _columns_for_text_search(Model)
+        if text_cols:
+            stmt = stmt.where(or_(*(col.ilike(t) for col in text_cols)))
+
+    if sort:
+        ui_sort = getattr(Model, "ui_sort", None)
+        if callable(ui_sort):
+            stmt = cast(Select[Any], ui_sort(stmt, sort, direction))
+        else:
+            col = getattr(Model, sort, None)
+            if col is not None:
+                stmt = stmt.order_by(sa_desc(col) if direction == "desc" else sa_asc(col))
+    else:
+        order_cols = getattr(Model, "ui_order_cols", ())
+        if order_cols:
+            for colname in order_cols:
+                col = getattr(Model, colname, None)
+                if col is not None:
+                    stmt = stmt.order_by(sa_asc(col))
+
+    if offset:
+        stmt = stmt.offset(offset)
+    if limit > 0:
+        stmt = stmt.limit(limit)
+
+    opts_attr = getattr(Model, "ui_eagerload", ())
+
+    opts: Iterable[Any]
+    if callable(opts_attr):
+        opts = cast(Iterable[Any], opts_attr())   # if you want, pass Model to it: opts_attr(Model)
+    else:
+        opts = cast(Iterable[Any], opts_attr)
+
+    for opt in opts:
+        stmt = stmt.options(opt)
+
+    return list(session.execute(stmt).scalars().all())
+
+def _resolve_column(Model, path: str):
+    """Return (selectable, joins:list[tuple[parent, attr]]) for 'col' or 'rel.col'"""
+    if '.' not in path:
+        col = _mapped_column(Model, path)
+        if col is None:
+            raise ValueError(f"Column '{path}' is not a mapped column on {Model.__name__}")
+        return col, []
+    rel_name, rel_field = path.split('.', 1)
+    rel_attr = getattr(Model, rel_name, None)
+    if getattr(rel_attr, 'property', None) is None:
+        raise ValueError(f"Column '{path}' is not a valid relationship on {Model.__name__}")
+    Rel = rel_attr.property.mapper.class_
+    col = _mapped_column(Rel, rel_field)
+    if col is None:
+        raise ValueError(f"Column '{path}' is not a mapped column on {Rel.__name__}")
+    return col, [(Model, rel_name)]
+
+def default_values(session, Model, *, id_: int, fields: Iterable[str]) -> dict[str, Any]:
+    fields = [f.strip() for f in fields if f.strip()]
+    if not fields:
+        raise ValueError("No fields provided for default_values")
+
+    mapper = inspect(Model)
+    pk = mapper.primary_key[0]
+
+    selects = []
+    joins = []
+    for f in fields:
+        col, j = _resolve_column(Model, f)
+        selects.append(col.label(f.replace('.', '_')))
+        joins.extend(j)
+
+    seen = set()
+    stmt = select(*selects).where(pk == id_)
+    current = Model
+    for parent, attr_name in joins:
+        key = (parent, attr_name)
+        if key in seen:
+            continue
+        seen.add(key)
+        stmt = stmt.join(getattr(parent, attr_name))
+
+    row = session.execute(stmt).one_or_none()
+    if row is None:
+        return {}
+
+    allow = getattr(Model, "ui_value_allow", None)
+    if allow:
+        for f in fields:
+            if f not in allow:
+                raise ValueError(f"Field '{f}' not allowed")
+
+    data = {}
+    for f in fields:
+        key = f.replace('.', '_')
+        data[f] = getattr(row, key, None)
+    return data
+
+def default_value(session, Model, *, id_: int, field: str) -> Any:
+    if '.' not in field:
+        col = _mapped_column(Model, field)
+        if col is None:
+            raise ValueError(f"Field '{field}' is not a mapped column on {Model.__name__}")
+        pk = inspect(Model).primary_key[0]
+        return session.scalar(select(col).where(pk == id_))
+
+    rel_name, rel_field = field.split('.', 1)
+    rel_attr = getattr(Model, rel_name, None)
+    if rel_attr is None or not hasattr(rel_attr, 'property'):
+        raise ValueError(f"Field '{field}' is not a valid relationship on {Model.__name__}")
+
+    Rel = rel_attr.property.mapper.class_
+    rel_col = _mapped_column(Rel, rel_field)
+    if rel_col is None:
+        raise ValueError(f"Field '{field}' is not a mapped column on {Rel.__name__}")
+
+    pk = inspect(Model).primary_key[0]
+    stmt = select(rel_col).join(getattr(Model, rel_name)).where(pk == id_).limit(1)
+    return session.scalar(stmt)
+
+def default_create(session, Model, payload):
+    label = infer_label_attr(Model)
+    obj = Model(**{label: payload.get(label) or payload.get("name")})
+    session.add(obj)
+    session.commit()
+    return obj
+
+def default_update(session, Model, id_, payload):
+    obj = session.get(Model, id_)
+    if not obj:
+        return None
+
+    editable = getattr(Model, 'ui_editable_cols', None)
+
+    changed = False
+    for k, v in payload.items():
+        if k == 'id':
+            continue
+
+        col = _mapped_column(Model, k)
+        if col is None:
+            continue
+
+        if editable and k not in editable:
+            continue
+
+        if v == '' or v is None:
+            nv = None
+        else:
+            try:
+                nv = int(v) if col.type.python_type is int else v
+            except Exception:
+                nv = v
+
+        setattr(obj, k, nv)
+        changed = True
+
+    if changed:
+        session.commit()
+    return obj
+
+def default_delete(session, Model, ids):
+    count = 0
+    for i in ids:
+        obj = session.get(Model, i)
+        if obj:
+            session.delete(obj); count += 1
+    session.commit()
+    return count
+
+def default_serialize(Model, obj, *, view=None):
+    # 1. Explicit config wins
+    label_attr = getattr(Model, 'ui_label_attr', None)
+
+    # 2. Otherwise, pick the first PREFERRED_LABELS that exists (can be @property or real column)
+    if not label_attr:
+        for candidate in PREFERRED_LABELS:
+            if hasattr(obj, candidate):
+                label_attr = candidate
+                break
+
+    # 3. Fallback to str(obj) if literally nothing found
+    if not label_attr:
+        name_val = str(obj)
+    else:
+        try:
+            name_val = getattr(obj, label_attr)
+        except Exception:
+            name_val = str(obj)
+
+    data = {'id': obj.id, 'name': name_val}
+
+    # Include extra attrs if defined
+    for attr in getattr(Model, 'ui_extra_attrs', ()):
+        if hasattr(obj, attr):
+            data[attr] = getattr(obj, attr)
+
+    return data