Skip to content

A smart, beautiful admin panel for Django

Zero config to start. One line to override anything. Install it, and every model you have is already administrable — with a data grid, charts, a command palette, light and dark themes and a schema map.

Get started View on GitHub

The Orders list in django-vantage: stat cards, a revenue chart, a status breakdown and a filtered data grid The Orders list in django-vantage, dark theme

Alpha — 0.1.0a1

Working and heavily tested — more than 3,600 tests — but the API may still move. See the FAQ for an honest account of what is and is not done. Use it on an internal tool; pin the exact version if you depend on it.

Sixty seconds

pip install --pre django-vantage

# settings.py
INSTALLED_APPS = ["vantage", ...]

# urls.py
from django.urls import path
from vantage import site

urlpatterns = [path("admin/", site.urls)]

That is the whole setup. Every model in the project is now administrable.

To change one model, write a panel.py next to its models.py:

from vantage import ModelPanel, register

from blog.models import Post


@register(Post)
class PostPanel(ModelPanel):
    list_display = ["title", "author", "status", "published_at"]  # (1)!
  1. Only the columns change. Filters, search fields and widgets stay inferred — partial configuration never opts you out of the rest.

Not sure what was inferred? Ask the package to write it down:

python manage.py vantage_scaffold blog

That prints a real panel.py containing exactly what Vantage decided. Own it, edit it, or delete an attribute to hand it back to inference. Full walkthrough: Getting started.

What you get

  • Inferred, attribute by attribute

    Columns, filters, search, sort order, widgets and query optimisation come from the model itself. Override one and the rest stay inferred.

    Inference

  • A real data grid

    Saved views, grouping, density, column picker, inline editing, trash and restore, bulk actions that scale to "select all 40,000 matching".

    Lists

  • Stats and charts in Python

    Stat and Chart declared on the panel, computed on the filtered rows, folded into as few queries as possible. Inline SVG — no JavaScript.

    Analytics

  • Forms with a layout DSL

    Tabs, Section and Row; editable inlines; a cascade-delete preview that shows the real tree of what goes with a record.

    Forms

  • Permissions that hold

    Three layers, enforced on every view, bulk action and API call. Read-only mode and impersonation live in the permission layer; multi-tenancy is fail-closed.

    Permissions

  • A JSON API for free

    Every panel, through the same permission and validation path as the UI, with cursor pagination and optimistic concurrency.

    The JSON API

  • One colour, a whole theme

    A brand colour becomes a complete light and dark palette with contrast checked. RTL from day one.

    Theming

  • Configurable from the UI too

    Columns, filters, stats, the dashboard, the sidebar and the branding — without a deploy, in one fixed order with panel.py.

    How configuration resolves

A closer look

Dashboard Dashboard

Command palette Command palette

Edit form Edit form

Catalogue list Catalogue list

Schema map Schema map

The two rules

Everything in this package follows from two rules.

1. Smart. Install it, and every model in your project already has a usable admin. Columns, filters, search fields, sort order, widgets and query optimisation are all inferred from the model itself. You write no configuration to get something worth looking at.

2. Easy. Every inferred default is overridable in one line, and overriding one thing overrides only that thing. Setting list_display changes the columns; your filters, search and widgets stay inferred. Partial configuration never opts you out of the rest of the intelligence.

The second rule is the load-bearing one. A tool where "customise anything" means "customise everything" gets abandoned the first time someone needs one different column. Inference is therefore resolved per attribute, not as a single all-or-nothing fallback — see Inference.

A developer's API is not the only person who has to configure this. Whatever you can set in panel.py or VANTAGE, the person running the panel day to day can often set from the UI instead — a list's columns and filters, the branding, the behaviour — without a deploy, and without your Python file becoming a fiction of what actually renders. The two ways compose in one fixed order, stated once: see How configuration resolves.

Requirements

Python 3.12+ and Django 5.2+ (including 6.x). Django is the only runtime dependency. No Node.js — the CSS and JavaScript ship prebuilt and depend on nothing. django-guardian, django-rules, django-mptt, django-treebeard and django-simple-history are all used when present and never required.

Map of the documentation

Page What is in it
Getting started Install, mount, first look, vantage_scaffold
How configuration resolves The one order everything composes in: UI override → panel.py/VANTAGE → inference
Panels The complete ModelPanel surface, attribute by attribute
Inference Exactly what is inferred and how, including the scoring rules
Lists Columns, filters, search, sorting, saved views, the per-list customisation rail, trash and restore, bulk actions, background jobs, export and import
Analytics Stat, Chart, the summary bar and the query-cost guarantees
Forms Widget inference, the layout DSL, editable inlines, readonly fields, validation, delete preview
Permissions The three-layer model, bulk row scoping, guardian, read-only, impersonation
The JSON API Read/write endpoints over every panel, cursor pagination, concurrency, errors as data
Multi-tenancy Scoping every registered model to a tenant, fail-closed, over both the panel and the API
Theming BRAND_COLOR, the derived palette, tokens, light/dark, RTL, custom CSS, the settings screen
Internationalization The shipped en/ar catalogues, activating a language, adding your own locale
Pages and dashboards Screens that are not models, dashboard widgets, and arranging cards from the UI
Settings Every VANTAGE key, its default and its effect
Deployment What to check before pointing this at production
Security The threat model, the deployment checks, upload screening, rate limiting, CSP
Accessibility What is guaranteed, what you must supply, what is still broken
Browser tests The Playwright layer over what a test client cannot see: focus, console errors, keyboard behaviour
Extending Slots, plugins, and adopting an existing admin.py
FAQ How it differs from django.contrib.admin, what it does not do

Licence and attribution

django-vantage is open source under the MIT licence, copyright © 2026 Mahmoud Ahmed. You may use, modify and redistribute it, commercially or not, provided the copyright notice and licence text are kept in every copy or substantial portion of the code. A link back to the project is always appreciated.