RAG & Knowledge Systems

STACKIT RAG Template

stackitcloud/rag-template

A template for building AI chatbots and document management systems using Retrieval-Augmented Generation (RAG), vector search, and FastAPI, designed for deployment on Kubernetes.

Open official repository ↗ Project homepage ↗
★ 86Stars
⑂ 10Forks
45Open issues
PythonLanguage
Apache-2.0License
Q@project.QualityScoreEditorial score

Project screenshots

Screenshot of STACKIT RAG Template Screenshot of STACKIT RAG Template

Overview

The STACKIT RAG Template provides a comprehensive architecture for developing and deploying AI-driven applications. It includes multiple microservices for document extraction, administration, and RAG-backed chatting, alongside a frontend interface. The system integrates with vector databases for semantic search, S3 for object storage, and Langfuse for monitoring. It supports both local development environments using Tilt and k3d, and production deployments on Kubernetes clusters.

Key features

  • Document management with automatic fallback chain across Docling, MarkItDown, and custom extractors
  • Support for multiple LLM and embedder providers (STACKIT, Ollama)
  • Model Context Protocol (MCP) server for standardized AI assistant integration
  • Langfuse integration for tracing, monitoring, and analytics
  • S3 object storage integration
  • Basic authentication for secure access
  • Kubernetes-ready deployment with Helm charts
  • Local development setup using Tilt and k3d with debug mode support

Requirements, installation and quick start

1. Install prerequisites: k3d, helm, and tilt.
2. Clone the repository.
3. Copy `.env.template` to `.env` and fill in the required values (Basic Auth, S3 credentials, Langfuse keys, optional LLM API keys).
4. Build the helm chart dependencies: `cd infrastructure/rag; helm dependency update; cd ../..`
5. Start the local cluster using Tilt: `tilt up` (for production mode) or `tilt up -- --dev=true` (for development mode with live code updates).

Usage

After deployment, access the WebUI via the configured ingress. Use the Admin App interface to upload documents or connect Confluence spaces. Once documents are processed and stored in the vector database, use the Chat App interface to ask questions about your documents. For debugging, start Tilt with `tilt up -- --debug=true --dev=true` and attach a debugger like VS Code with debugpy on the specified ports.

Model compatibility and use cases

Supports OpenAI-compatible LLMs via the STACKIT model serving service and Ollama as an LLM provider.

License and risk notes

Apache-2.0

Editorial verification 2026-08-02: repository URL, owner, description, license and repository statistics were reviewed. License metadata: Apache-2.0. README was fetched for the channel draft; re-check repository dependencies, releases and model terms before production use.

Release and maintenance

Not stated in the repository metadata

LanceDB

lancedb/lancedb

★ 11.1KRust

EvalScope

modelscope/evalscope

★ 3.2KPython

Infino

infino-ai/infino

★ 67Rust