kendr · Super-RAG

Select a knowledge base

Create or select a KB from the left panel to get started.

📂 Sources

🗄 Vector DB

🎯 Reranker

🤖 Agents

🔍 Query / Test

Select or create a knowledge base to manage sources.

Add Source

Source Type

Path on this machine

Max files

Extensions (comma-sep, blank=all)

Recursive

📁

Drop files here or click to upload

PDF, DOCX, TXT, MD, XLSX, CSV, PPTX supported

URL

Max pages to crawl

Same domain only

Database URL

Tables (comma-sep, blank=all)

Schema (optional)

OneDrive path (blank = root)

Requires Microsoft integration to be configured in Setup.

Label (optional)

Indexed Sources

No sources yet.

Vector Store Backend

Backend

ChromaDB runs locally with no setup. Qdrant and pgvector support remote deployments.

ChromaDB persist path

Qdrant URL

API Key (Qdrant Cloud)

PostgreSQL URL (with pgvector)

Embedding Model

OpenAI embeddings require OPENAI_API_KEY to be configured.

Backend Status

Load a KB to see status.

Reranking Algorithm

Algorithm

Top-K results to return

Fetch K (candidates before reranking)

Fetch this many from the vector store, then rerank down to Top-K.

Keyword weight (0 = pure vector, 1 = pure keyword) 0.3

Cross-encoder model

Install sentence-transformers: pip install sentence-transformers

Cohere API Key

Agent Connections (Super-RAG)

Enable agents to automatically query this knowledge base when answering tasks. This transforms kendr into a Super-RAG — every enabled agent can retrieve context from this KB.

Active KB used by agents

Enabled agents will use the active KB (marked with ★ in the left panel). Switch the active KB to change which knowledge base agents access.

AI Answer

Retrieved Chunks