Chat Demo: https://chat.daniel-lambrecht-dev.de/

You need a passwort to access the demo of the bot.

OpsPilot

OpsPilot is a Rasa Chatbot designed to generate dockerfiles, K8s Manifests and Pipeline Templates intelligent, conversation-based chatbot to support developers in DevOps tasks - self-hosted, extensible and specialized in infrastructure-as-code, CI/CD and containerization.

This project uses Poetry for dependency management and Rasa for development for the NLP chatbot.

---

📦 Installation

install dependencies:

poetry install --no-root

(optional) add dependency

poetry add <package name>

---

🐍 Activate environment (optional)

poetry run <command>
# or with poetry plugin
poetry self add poetry-plugin-shell
poetry shell

Train the Model

chmod +x ./scripts/train_model.sh 
./scripts/train_model.sh 

!!! Run Locally with Docker

'to Run the bot locally just run:

docker-compose up -d --build

Start up all services. This activates the following:

• rasa-chatbot
• rasa-chatbot-frontend
• rasa-action-server

It can take a while to start up the first time as all the images have to be drawn and built first.

After starting, you can navigate to at http://localhost:8080

To exit the environment simply

docker-compose down

Login data of the front end

user: dev
password: sose2025

Performing locally with poetry

Rasa API Launch

poetry run rasa run --enable-api --cors "*" --debug

Rasa Action server launch

poetry run rasa run actions

launch frontend

cd frontend && yarn dev

---

🚀 Rasa + Poetry commands (Cheat Sheet)

Rasa Command	Beschreibung	Mit Poetry ausführen
rasa init	Erstellt ein neues Projekt mit Beispieldaten	poetry run rasa init
rasa train	Trainiert ein Modell und speichert es in ./models	poetry run rasa train
rasa interactive	Interaktive Trainingssession	poetry run rasa interactive
rasa shell	Öffnet eine CLI-Chat-Session mit deinem Bot	poetry run rasa shell
rasa run	Startet den Rasa-Server	poetry run rasa run
rasa run actions	Startet den Action-Server mit dem Rasa SDK	poetry run rasa run actions
rasa visualize	Generiert eine visuelle Darstellung deiner Stories	poetry run rasa visualize
rasa test	Führt Tests auf einem trainierten Modell aus	poetry run rasa test
rasa test e2e	Führt End-to-End-Tests inklusive Actions durch	poetry run rasa test e2e
rasa data split nlu	Teilt die NLU-Daten im Verhältnis 80/20	poetry run rasa data split nlu
rasa data split stories	Teilt die Story-Daten im Verhältnis 80/20	poetry run rasa data split stories
rasa data convert	Konvertiert Trainingsdaten in andere Formate	poetry run rasa data convert
rasa data migrate	Migriert Domain-Dateien von Rasa 2.x zu 3.x	poetry run rasa data migrate
rasa data validate	Prüft Domain-, NLU- und Story-Daten auf Konsistenz	poetry run rasa data validate
rasa export	Exportiert Konversationen aus dem Tracker Store	poetry run rasa export
rasa evaluate markers	Extrahiert Marker aus dem Tracker Store	poetry run rasa evaluate markers
rasa marker upload	Lädt Marker-Konfigurationen zu Analytics hoch	poetry run rasa marker upload
rasa license	Zeigt Lizenzinformationen an	poetry run rasa license
rasa -h	Listet alle verfügbaren Rasa-Kommandos auf	poetry run rasa -h

🎯 Goal of the Project

The aim of this project is to develop a specialized, contextual chatbot that:

• primarily:
  • Generate Dockerfile templates (customized for language, tools/technologies, build type)
  • Create CI/CD pipeline templates for GitHub Actions or GitLab CI (or both depending on how much time)
  • conducts structured dialogs to query user-specific requirements
  • runs locally, offline and completely in your own cluster if desired
• optional:
  • kubernetes deployment templates (YAML)

---

🛠️ Technology-Stack

Komponente	Technologie
Bot-Framework	Rasa Open Source
Template Engine	Jinja2 (Dockerfiles, YAML etc.)
Backend-Logik	Python (Custom Actions)
Frontend	React Typescript frontend powered by Vite and bootstrap for styling
Deployment	Kubernetes (K3s), Helm, Traefik or local with docker
CI/CD	GitHub Actions, GitLab CI

---

🤖 Example Dialoge (Prototype)

User: I need a Dockerfile Bot: yarn or npm? User: npm
Bot: image version? User: 20 Bot: ✅ here is your dockerfile:

---