If you've ever tried to integrate stateful, streaming AI agents into a Django project, you know the pain: managing state, handling Server-Sent Events (SSE), securing database tools, and preventing context window overflows.
That's why I'm excited to announce django-langgraph-agent, an open-source, production-ready Django package for building streaming AI agents powered by LangGraph.
We didn't just build this in a vacuum. This package is already battle-tested in production, powering intelligent features at Tathbeet (our Quran memorization platform serving thousands of users). We built it to solve real-world problems, and now we're sharing it with the community.
Why you'll love it:
- No-Code Agent Tuning: Create, tweak, and update agent prompts and configurations directly from the Django Admin UI. No redeployments required.
- Zero-Boilerplate Streaming: Built-in SSE endpoints for real-time chat and tool execution out of the box.
- Human-in-the-Loop: Automatically pause agents on sensitive actions (like database writes) and resume only upon user confirmation.
- Safe Auto-ORM Tools: Secure, read/write database operations are automatically attached via a simple, strict MODEL_WHITELIST in your settings.
- Resilient by Design: Multi-model fallback support and auto-summarization to keep your context windows in check.
Get started in under 3 minutes:
- pip install django-langgraph-agent
- Add it to your INSTALLED_APPS and urls.py
- Configure your API key and model whitelist in settings.py
- Create your first agent in the Django Admin and you're live!
We built this to make Django the absolute best framework for pragmatic, production-grade AI applications.
Check out the repository, read the docs, and drop a star if you find it useful:
https://github.com/bixat/django-langgraph-agent
I'd love to hear your thoughts, feedback, or see the cool agents you build with it! Let's push the boundaries of what Django can do in the AI era.
Top comments (0)