pgkiln

Website · Documentation · Tutorial · Coming from Oracle APEX · Roadmap

A low-code, SQL-driven application builder for PostgreSQL, modeled on Oracle APEX. Applications are data: pages, regions, items, buttons, dynamic actions, validations and processes are rows in the meta schema. A runtime turns those rows into web pages, and the builder is an editor for them.

pgkiln is an independent open source project and is not affiliated with Oracle. Oracle and APEX are trademarks of Oracle.

Screenshots

Everything below is the HR example application (npm run example:hr) and the builder. The screenshots are made by npm run screenshots.

Dashboard: metric cards and charts, all of it SQL	Interactive report: search, filters, sort, download
Smart filters and faceted search	Map and cards regions on the same table
Charts, in dark mode	Calendar of leave requests
App Builder home	SQL Workshop: object browser with RLS and grants

Quick start

With Docker (no Node.js needed):

mkdir pgkiln && cd pgkiln
curl -fsSLO https://raw.githubusercontent.com/pgkiln/pgkiln/main/deploy/compose.yaml
curl -fsSL -o .env https://raw.githubusercontent.com/pgkiln/pgkiln/main/deploy/.env.example   # fill in the four secrets
docker compose up -d   # http://127.0.0.1:3100/builder (admin / the password from .env)

Or from the source, for developing:

npm install
cp .env.example .env
npm run setup        # Postgres 17 on localhost:5434 + the pgkiln schema (migrations)
npm run dev          # http://127.0.0.1:3100/builder (admin / admin)

Then build an application on your own tables: Create application in the builder generates report and form pages from a table, and everything else (regions, items, processes, security, translations, documents, approvals) is added in the page designer and Shared Components. The tutorial walks through one.

The example application

pgkiln itself contains no application. The repository ships an example built on it, HR (examples/hr/): employees and departments, leave requests with approvals, a dashboard, a REST API, translations, documents, automations and row level security, all of it ordinary pgkiln metadata and PostgreSQL code, the way you would build your own.

npm run example:hr   # http://127.0.0.1:3100/a/hr

Users (password = username): king is admin and manager (the president), blake and jones are managers, allen and scott are employees, and demo is an admin who is not an employee. Sign in as different users to see authorization schemes and row level security at work.

Developing pgkiln

npm test             # unit + security regression tests (installs the HR example as their fixture)
npm run test:e2e     # browser tests on phone/tablet/desktop (npx playwright install chromium first)

npm run db:reset recreates the database (pgkiln only). Schema changes go in a new db/migrations/NNN_*.sql file; npm run db:migrate applies pending ones.

Working with AI coding agents

pgkiln ships an MCP server (pgkiln mcp), so Claude Code, Cursor and other agents can find an application, read and explain its pages, look at the tables and RLS policies behind them, search this documentation, and change the application: export it as files (one YAML file per component, SQL inline), edit them, show the diff, and import them again in one transaction. In a checkout, .mcp.json starts it for Claude Code; for your own project:

claude mcp add pgkiln -- /path/to/pgkiln/bin/pgkiln.js mcp

Without a checkout, npx -y pgkiln mcp (with DATABASE_URL set) or, for the Docker installation, docker compose exec -T pgkiln node bin/pgkiln.js mcp starts the same server.

Then ask things like "explain what happens when I press Save on page 3 of hr" or "add a hire date column to the employee report and show me the diff". Queries it runs are read-only. See AI coding agents.

Without an agent, the same export and import is on the command line (pgkiln export hr --format text, pgkiln diff, pgkiln import --replace, see the CLI) and in the builder (Export / Import).

Documentation

What's next: ROADMAP.md.

The user guide explains how everything works: installation and configuration, concepts, the builder, every region and item type, processing, dynamic actions, security, the SQL reference, a step-by-step tutorial, a guide for Oracle APEX developers, and developing pgkiln itself.

What's in the box

For a full comparison with Oracle APEX 26.2, including what's missing, see docs/apex-feature-parity.md.

This is the beginning of the README. Read the rest at the original. Read the full README ↗