APIJSON Review: 18K Stars, Zero-Code ORM Where the Backend Writes No Endpoints

When I first read APIJSON’s tagline—“zero-code APIs and docs”—my eyes rolled so hard they almost fell out. It sounded like marketing fluff from a vaporware product. Then I noticed it’s open-sourced by Tencent and has over 18,000 stars. I figured it was worth half a day to test.

What This Project Does

In short, APIJSON is a Java ORM library with a radical premise: the frontend describes what data it wants using JSON, and the backend doesn’t need to write Controllers, Services, or even SQL. The framework translates the JSON into queries and returns results automatically.

It sounds extreme, but the query capabilities are surprisingly comprehensive: single-table lookups, multi-table joins, aggregations, subqueries, even complex nesting. And the auto-generated docs stay in sync with the code—no more maintaining Swagger by hand.

The project has been around since 2016, nearly a decade of maintenance, which says something about its staying power.

Core Features

Zero-Code CRUD

This is the headline feature. After deploying the APIJSON backend, the frontend sends JSON like this:

{
  "User": {
    "id": 1
  }
}

And gets back the user with id 1. Want to join tables? Just add fields:

{
  "User": {
    "id": 1,
    "Moment[]": {
      "Moment": {
        "userId@": "/User/id"
      }
    }
  }
}

The syntax takes some learning, but once you’re fluent it’s faster than writing SQL.

Auto-Generated Docs

Every time the schema changes, the API docs update automatically. No more “the endpoint changed but the docs didn’t” embarrassments. The generated docs look similar to Swagger and can be tested directly in the browser.

Multi-Database Support

MySQL, PostgreSQL, Oracle, SQL Server, ClickHouse, TiDB, Snowflake—over twenty databases are supported. I tested MySQL and PostgreSQL and didn’t run into compatibility issues.

Security Controls

It has its own permission model: which tables are queryable, which fields can be returned, even max result limits per query. Given that “the frontend queries the database directly,” these safety guardrails are essential.

Real-World Use Cases

Admin dashboards: Projects where 80% of endpoints are CRUD operations. APIJSON eliminates mountains of boilerplate.

Prototype validation: When you need a working backend for a product idea fast, APIJSON can get you a complete API in hours.

Data dashboards: Combined with its aggregation queries, the frontend can request stats directly without backend report endpoints.

Quick Start

Clone the demo project:

git clone https://github.com/Tencent/APIJSON.git

Build with Maven, configure your database connection, and you’re running. The official docs include a Spring Boot integration guide—about 10 minutes to your first working endpoint.

Pros and Cons

Pros:

• Genuinely reduces massive amounts of repetitive backend code
• Auto-generated docs with near-zero maintenance cost
• Query capabilities exceed expectations, handling complex joins
• Extensive multi-database support
• Tencent backing with an active community

Cons:

• Steep learning curve; the JSON query syntax takes time to get used to
• Debugging is painful; JSON errors sometimes produce unhelpful messages
• Performance can suffer on complex queries that generate inefficient SQL
• Team adoption is tricky—many backend devs resist the “frontend commands the backend” model
• I’ve only used it on small projects; haven’t personally verified behavior at large production scale

Who Should Use It

• Indie developers who need to prototype fast
• Teams building lots of admin panels
• Projects with limited backend resources where frontend devs need to handle API work
• Java projects looking to cut boilerplate

Apache-2.0 licensed.

---

About the Author

Liudingyu is a full-stack developer and heavy GitHub user. With 900+ starred repos over the past 3 years, this site only covers tools I’ve actually used or deeply researched.

📧 Found a great tool to recommend? Email [email protected]