Saison 2026: täglich aktualisiert 💡 Eigene App, Webseite, KI-Agenten oder Digital-Idee? Das Team hinter BreisgauStraussen entwickelt sie für dich!💡 Eigene App oder KI-Agenten? Wir entwickeln sie für dich!
BreisgauStraussen.deDein Straußwirtschaftenführer für Südbaden

neooutput

Agentic coding with Python and Django: a workshop for web and backend teams

Python is dynamically typed, and Django projects rely on conventions that are not always visible in the code. Coding agents therefore need more context than they would in Java or C#. In the neooutput workshop your team learns to describe environments, dependencies and test commands precisely, add type hints where they help most and guide agents so migrations, views and serializers stay consistent with each other.

Book a call for freeneooutput

Deutsche Version

In short

An agentic coding workshop for Python and Django teams shows how coding agents such as Claude Code, OpenAI Codex or Cursor implement features, tests and migrations in existing Django projects. neooutput focuses on reproducible environments, pytest as the final judge and rules written in CLAUDE.md or AGENTS.md, so results stay consistent whichever tool you choose.

Your benefits

Tool-agnostic method

The approach works with Claude Code, Codex and Cursor, so your team is not locked into one vendor.

pytest as referee

Every agent change is checked against the test suite, and missing tests are generated first.

Clean migrations

The agent produces traceable migrations, and conflicts between branches surface early.

Less lost context

Knowledge about settings, apps and Celery tasks lives in files any agent can read.

Why Python needs extra care with agents

In Python an agent misses many signals a compiler would give in Java or C#. Errors appear at runtime or in tests. A reliable test suite is therefore the single most important requirement for productive agentic coding in Django.

Type hints checked by mypy or Pyright and a linter such as Ruff help further. A hook that runs linting and tests after every edit gives the agent immediate feedback and saves your team manual correction loops.

Migrations deserve special attention. When several agents or people work on models in parallel, competing migration files appear quickly. In the workshop we agree who creates migrations, how conflicts are resolved and that the agent never rewrites migrations that have already been deployed.

Four stages of the workshop

The structure is proven but adapted to your projects, your tool and the team's experience. The goal is that every participant finishes at least one real task with an agent and reviews it. You also keep reusable building blocks: a rules file, a hook configuration for Ruff and pytest, and a short review checklist stored directly in the repository.

  1. neooutput's AI code engineers review project layout, dependency management with pip, Poetry or uv, and the existing tests.
  2. Together we write a rules file, CLAUDE.md for Claude Code or AGENTS.md for Codex, covering commands, conventions and things the agent must never do.
  3. The team works on real tasks, for example a new Django model with admin, API endpoint and tests.
  4. Finally we agree which tasks agents will own from now on and how reviews will work.

Data and knowledge around the code

Python teams often work close to data: exports, fixtures, log files. If these contain personal information, it should be anonymised before it goes into a prompt. Textwash Pro does this automatically and supports GDPR-compliant workflows.

If you also want specialist documents, manuals or policies to be accessible through AI, kalAIDoCreate offers a platform for knowledge assistants and document AI for companies and public sector organisations.

Frequently asked questions

Which tool do you recommend for Django?

It depends on your team. Claude Code is strong at multi-file work in the terminal, OpenAI Codex adds a cloud mode for running tasks in parallel next to its CLI, and Cursor embeds the agent in the editor. We compare them on your code during the workshop.

Does the workshop also cover FastAPI or Flask?

Yes. The core principles apply to any Python project. The Django focus mainly concerns the ORM, migrations and admin; with FastAPI or Flask the emphasis shifts to Pydantic models, routing and tests.

Is the workshop remote or on site?

Both work. neooutput operates remotely across the DACH region and internationally, and also on site at your offices. Hands-on work on the code is equally effective in either format.