Skip to main content

Welcome to ONQL

ONQL stands for Object Notation Query Language. It's a context-aware query language that lets you describe the shape of the data you want — and skips the API layer entirely.

shop.orders[status = "paid"]{
  id, total, placed_at,
  customer{name, email},
  items{product{name}, qty, price}
}

That single query is a SELECT, four JOINs, an authorization rule, and a JSON serializer — all in one expression. It can be sent directly from your frontend.

Why ONQL exists​

Most backends spend 60% of their engineering time writing the same code:

  1. Receive an HTTP request.
  2. Authenticate the caller.
  3. Build a SQL query that joins three tables.
  4. Filter by user_id so people only see their own data.
  5. Serialize the result into JSON.
  6. Repeat for the next screen.

ONQL deletes that loop. You define your tables, relationships, and authorization rules once in a small protocol file, and any client can issue queries directly. The query language understands joins, paging, sorting, aggregates, and — uniquely — context-aware row scoping.

What makes it different​

  • Shape-first. Your query and your response have the same structure.
  • Context-aware. Same query → different rows depending on who is asking. No middleware required.
  • Zero glue code. One protocol file replaces an entire BFF / API layer.
  • Declarative JSON migrations. Describe the schema you want; ONQL diffs and applies it. No CREATE TABLE, no migration files.
  • Built on BadgerDB. Embedded LSM-tree storage with reverse indexes on every column. RAM buffering, async disk persistence, single binary.
  • Open source. Apache 2.0.

Where to go next​

If you want to…Read this
Run ONQL on your machineGetting Started → Installation
Define tables without writing SQLGetting Started → Schema & Migrations
Learn the language in 5 minutesTour → Basics
See real production-style queriesCookbook → Dashboards
Look up an operator or functionReference → Cheat Sheet
Understand how queries are executedConcepts → Query Execution

A note on the name​

A query should look like the JSON you want back. No SELECTs, no JOINs, no boilerplate — just the shape of your data.

users{id, name, orders{total}}
{
  "users": [
    { "id": 1, "name": "Ada", "orders": [{ "total": 99 }] }
  ]
}

Query in. JSON out. Same shape. That's ONQL.