Overview
Enjab RAG is the shared knowledge brain for Enjab's agents, one place that ingests, processes, and serves operational knowledge over MCP and a REST API.
Enjab RAG is the shared, long-term memory for every Enjab agent. You send it knowledge (files, text, images, audio, or links); it extracts and processes the content, embeds it, and serves it back as searchable context, over a native MCP server and a simple REST API.
- Home: rag.enjab.ae
- MCP endpoint:
https://rag.enjab.ae/api/mcp - REST base:
https://rag.enjab.ae
Every Enjab agent connects to Enjab RAG. It is the single source of shared operational knowledge (how the clinics run). Wiring each agent to it is what keeps them consistent, instead of every agent holding its own stale copy. If you are building or running an Enjab agent, connect it, see Connect an agent.
What it does
- Ingests anything — PDF, Word, spreadsheets, images, audio, text, and web links.
- Processes, not just stores — extracts the text (OCR for images, transcription for audio), then writes a title, summary, tags, and structured metadata with a model.
- Embeds and indexes — chunks the content and stores OpenAI
text-embedding-3-largevectors in Postgres (pgvector) for semantic search. - Serves retrieval — agents call
search(andfetch) to ground their answers in real, current operational knowledge.
The agents do the reasoning and answering; Enjab RAG is the retrieval layer they share.
Quickstart
Mint a key at rag.enjab.ae/dash/keys, then add knowledge and retrieve it:
# 1. Add knowledge (write key)
curl -s https://rag.enjab.ae/api/ingest \
-H "Authorization: Bearer enjabrag_WRITE_KEY" \
-H "Content-Type: application/json" \
-d '{"text":"The Dermatology clinic is open 8am to 9pm daily."}'
# 2. Retrieve it (read key)
curl -s https://rag.enjab.ae/api/search \
-H "Authorization: Bearer enjabrag_READ_KEY" \
-H "Content-Type: application/json" \
-d '{"query":"when is dermatology open?"}'Building an agent? Skip straight to Connect an agent, the MCP server is the native path, no HTTP plumbing.
Two ways to integrate
Connect an agent (MCP)
The native path. Point any MCP client at the server and it gets the search + fetch tools. This is how agents should connect.
REST API
POST /api/search and /api/ingest for backends and tools that are not MCP clients.
Authentication
API keys (Bearer), read vs write scopes, and where to mint them.
How it works
The ingestion pipeline, supported inputs, collections, and the embedding model.
Endpoints at a glance
| Method | Path | Purpose | Key |
|---|---|---|---|
POST | /api/mcp | MCP server: the search + fetch tools | any |
POST | /api/search | Semantic search, ranked passages | read |
POST | /api/ingest | Add a file, URL, or text | write |
All requests authenticate with an API key in the Authorization: Bearer header. See Authentication.
Coding agents: these docs are published for agents too. Point your agent at https://developers.enjab.ae/llms.txt plus this section and it has everything it needs to connect a project to Enjab RAG.