RAG Service

A complete RAG service with admin dashboard for managing your knowledge base.

Source code: github.com/zeabur/rag-service

Getting Started

Step 1: Create a Zeabur Project

Go to Zeabur Dashboard and create a new project.

Step 2: Deploy

Add RAG Service from the Marketplace and provide a Zeabur AI Hub API Key for answer generation plus a Gemini API Key for embeddings. The template creates a private PostgreSQL 17 service with pgvector and injects its connection string into the RAG service automatically.

Schema migrations run automatically on first startup — no manual setup needed.

Features

• Hybrid search: semantic (pgvector) + BM25 keyword with RRF fusion
• Streaming RAG answers via configurable LLM model
• Admin dashboard: chunk browser, query signals, reports, audit log
• API key management with scopes and expiry
• Knowledge learning API: add new chunks via /api/learn
• Feedback collection on query results

API Endpoints

Endpoint	Method	Auth	Description
/api/query	POST	API key	Search knowledge base + optional RAG answer
/api/learn	POST	API key	Add new knowledge chunks
/api/report	POST	API key	Report content issues
/api/feedback	POST	API key	Submit result feedback
/api/admin/*	GET/POST	API key (admin)	API key management, signals, reports, chunks
/dashboard	GET	Basic Auth	Admin dashboard

Quick Test

curl -X POST "https://your-domain/api/query" \
  -H "Authorization: Bearer $RAG_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query": "how to deploy", "mode": "hybrid", "top_k": 5}'

Agent Integration (Claude Code Plugin)

Install the plugin to let Claude Code agents search and contribute to your knowledge base:

claude plugins marketplace add zeabur/rag-service
claude plugins install rag-service@rag-service
claude
# Inside Claude Code, run: /zeabur-rag-setup

The plugin provides 8 skills: search, learn, report for everyday use, and triage, inspect, edit, curate for admin curation. Agents automatically search the KB when answering questions and can contribute new knowledge after solving problems.

Environment Variables

Variable	Required	Description
DATABASE_URL	Auto	Private PostgreSQL connection string
ZEABUR_AI_HUB_API_KEY	Yes	Zeabur AI Hub key for LLM inference
GEMINI_API_KEY	Yes	Gemini API key for embeddings
EMBEDDING_MODEL	No	Embedding model (default: gemini-embedding-2)
EMBEDDING_DIMENSIONS	No	pgvector dimension (default: 1536)
RAG_API_KEY	Auto	API key for service access (auto-generated)
RAG_BASIC_AUTH	Auto	Dashboard auth (auto-generated, format admin:password)
RAG_MODEL	No	LLM model (default: gemini-2.5-flash-lite)
CORS_ORIGIN	No	CORS origin (default: service URL)

Embeddings use gemini-embedding-2 (1536d) through the Gemini API by default.

Schema migrations run automatically on first startup.

License

MIT — Built by Zeabur