// rag Β· xuwei95/ezdata

ezdata

AI-native data platform: 60+ connectors, ETL, scheduling, and RAG search.

πŸ’Ž Hidden gemβ˜… 293 starsβ‘‚ 59 forksMITupdated 2026-08
β–¦ visual β€” this project's README ships no embeddable screenshot,
so here's a placeholder. The guide below is complete regardless.
// project facts

Project facts

ezdata is an open-source rag & knowledge project by xuwei95 (xuwei95/ezdata on GitHub). It has 293 GitHub stars and 59 forks, is released under the MIT license, and was last updated on GitHub on 14 August 2026.

Repository
xuwei95/ezdata
GitHub stars
293
Forks
59
License
MIT
Last updated on GitHub
14 August 2026
Listing
Hidden gem β€” a shorter page for a lesser-known project
Official site
Official site
// how it works

How it works

  1. Clone the repository

    Download the source and enter the folder.

  2. Install the platform

    Pull the Python dependencies.

  3. Connect a source

    Add one of the 60+ connectors per the README.

  4. Ask your data

    Let the agents query, chart, and analyze.

// features

Why people love it

60+ connectors

Ingest from databases, APIs, and files.

dlt-based ETL

Reliable extraction and loading pipelines.

DAG scheduling

Orchestrate recurring data jobs.

Per-source RAG

Knowledge bases scoped to each source.

Sandboxed AI agents

Agents query and chart safely.

Text-to-SQL

Ask questions in plain language.

Advertisement
// run it

Run it yourself

  1. Step 1 β€” Clone the repo

    Download the source and move into the project folder.

    git clone https://github.com/xuwei95/ezdata.git
    cd ezdata
  2. Step 2 β€” Install dependencies

    Pull the Python packages it needs.

    pip install -r requirements.txt
  3. Step 3 β€” Launch the platform

    Start the services per the README.

    docker compose up -d
// use cases

What you can build

Advertisement
// faq

Questions, answered

Who maintains ezdata?

It is an open-source project at xuwei95/ezdata with 59 forks.

Is it free for commercial use?

Yes. It is MIT licensed.

How do the AI agents query data?

Through sandboxed agents with text-to-SQL over your connected sources.

Which sources are supported?

60+ connectors spanning databases, APIs, and files β€” see the README.

Do I need to self-host?

Yes. It is designed to run on your own infrastructure.

// keep exploring

Similar tools

RAG & Knowledge

ragflow

RAGFlow is a leading open-source Retrieval-Augmented Generation (RAG) engine that fuses cutting-edge RAG with Agent capabilities to create a superior context…

β˜… 91.6k starsApache-2.0
View guide β†’
Automation

actor-rag-web-browser

Apify actor that feeds RAG pipelines with fresh scraped web text.

β˜… 80 starsApache-2.0
View guide β†’
RAG & Knowledge

rag-template

Run rag-template locally: template for ai chatbots & document management using re.

β˜… 87 starsApache-2.0
View guide β†’
RAG & Knowledge

seekdb

seekdb β€” the AI-native search database unifying vector, text and structured data for agents.

β˜… 3.1k starsApache-2.0
View guide β†’
RAG & Knowledge

swirl-search

Federated AI search and RAG across 100+ apps β€” answers with sources, without moving your data

β˜… 3.0k starsApache-2.0
View guide β†’
RAG & Knowledge

trieve

All-in-one API platform for search, recommendations, RAG and analytics β€” self-hostable

β˜… 2.7k starsMIT
View guide β†’
// alternatives

Alternatives in RAG & Knowledge

The most-starred other RAG & Knowledge projects in the LibreAI catalogue β€” the realistic shortlist if ezdata is not the right fit.

Get ezdata

Free and open β€” grab the source, star the repo, and start creating.