django-admin-ai-assistant 0.1.2

Embedded AI assistant for Django Admin with read-only ORM execution.

django-admin-ai-assistant

Reusable Django application with an embedded AI assistant inside Django Admin.

Installation

1. Install package dependencies and the app:

pip install -e .

2. Add apps in settings.py:

INSTALLED_APPS = [
    # ...
    "rest_framework",
    "django_ai_admin",
]

3. Add URLs in root urls.py:

from django.urls import include, path

urlpatterns = [
    # ...
    path("ai-assistant/", include("django_ai_admin.urls")),
]

4. Run migrations:

python manage.py migrate

5. Optional settings in settings.py:

DJANGO_AI_ADMIN_URL_PREFIX = "ai-assistant"
DJANGO_AI_ADMIN_ADMIN_SITE = "backend.admin_site.admin_site"  # optional
DJANGO_AI_ADMIN_OPENAI_BASE_URL = "https://api.openai.com/v1"

Usage

1) Configure AI in Django Admin

1. Log in as superuser.
2. Open Admin -> Django Ai Admin -> AI configs.
3. Create or edit the config record and fill:
   • provider: openai
   • api_key: your API key
   • model: for example gpt-4o-mini
   • temperature: for example 0.2
   • max_tokens: for example 1024
   • timeout_sec: for example 30
4. Save.

2) Use the assistant

1. Open any Django Admin page as a staff user.
2. Click Open AI in the admin header.
3. Start a new chat and ask a data question about your project models.
4. Review response, result, and optional details (interpretation/explanation/code).

Docker Demo Project

A ready-to-run example project is included in example_project/ with simple polls models (Question, Choice) and admin integration.

From the repository root:

cd example_project
make dev

Then open:

• http://localhost:8000/admin/
• login: admin / admin

See example_project/README.md for details.

Release (PyPI)

Package metadata currently uses:

• package name: django-admin-ai-assistant
• version: 0.1.0

To publish from GitHub Actions:

1. Create a PyPI API token and add it to repository secrets as PYPI_API_TOKEN.
2. Make sure pyproject.toml version matches the release (for example 0.1.0).
3. Create and push a tag:

git tag v0.1.0
git push origin v0.1.0

What happens automatically:

• Release workflow builds sdist and wheel
• creates a GitHub Release with attached artifacts
• publishes to PyPI (if PYPI_API_TOKEN is configured)

Install command after publish:

pip install django-admin-ai-assistant==0.1.0