Copyright (c) 2025 SoftwareTree, Inc. All Rights Reserved.

Make Your Database AI-Ready — Automatically

Last updated: 2026-08-19 1:44 AM PDT

Your business data is already in databases. The missing piece is making it accessible to AI.

ORM_Skyway automation tool connects your existing relational databases to AI agents — automatically. Point it at your database, and within minutes you have a secure, governed API layer that lets AI reason about your business objects: customers, orders, products, employees — whatever your domain holds.

Immediately start leveraging your data for AI applications through a secure and efficient ORM pipeline.

No new database. No data migration. No custom API development. And because you control exactly what the AI can see, sensitive data stays protected.

Turn your existing database into an AI-ready asset — in minutes, not months.

---

For developers: Your relational database. AI-ready in one command.

ORM_Skyway reverse-engineers any existing relational database into a curated JSON object model (using JDX), packages it as a RESTful microservice (using Gilhari), and connects it to AI agents via MCP (using ORMCP) — automatically. Instead of raw SQL or generic REST, AI agents get a curated, object-oriented view of your domain data. Shaped JSON objects keep token usage low. Sensitive columns removed from the ORM spec stay invisible to the agent. The .jdx mapping file is your governance boundary.

MySQL, PostgreSQL, Oracle, SQLite, DB2, SQL Server, ... — one tool, one command, zero hand-written API code.

---

What it does

ORM_Skyway (orm_skyway.py) automates the journey from a relational database to a live REST microservice (Gilhari), and from there to an AI agent that can query your data through natural language.

Your Database  ──►  Object Model  ──►  REST Microservice  ──►  AI Agent
                   (auto-generated)     (Docker image)        (via ORMCP)

The AI agent gets a curated, object-oriented view of your data — not raw tables or SQL. This improves reasoning clarity, keeps token usage low, and gives you a clean governance boundary: the agent sees only what your domain model exposes.

The ORM_Skyway Pipeline

ORM_Skyway creates and connects three elevated ORM pipelines — like a skyway connecting buildings above ground level:

ORMCP Pipeline   ─────────────────────────────────  ← AI / MCP layer
                              ↑
Gilhari Pipeline ─────────────────────────────────  ← REST microservice layer
                              ↑
JDX Pipeline     ─────────────────────────────────  ← Java/JSON ORM layer
                              ↑
Your Database    ═════════════════════════════════  ← foundation

Each layer builds on the one below. ORM_Skyway automates the entire stack — from raw database schema to AI-ready API — in one elevated, protected pathway.

---

The workflow at a glance

Phase	What happens	How
1 — Reverse engineer	Reads your DB schema, generates Java/JSON model classes and ORM spec	Script (automated)
2 — Refine & Curate	Rename attributes, hide sensitive columns, curate the model; verify mapping with JDXDemo GUI tool	Manual — optional but recommended (edit one text file)
3 — Package	Builds a Gilhari microservice Docker image with REST APIs for every mapped class	Script (automated)
4 — Run & test	Start the container; verify with curl or Postman	Manual — optional but recommended
5 — Connect AI	Point ORMCP at the running Gilhari microservice; add one config snippet to your AI client	Minutes

---

Prerequisites

• Python 3.8+
• JDK 8+ on your PATH
• Gilhari SDK (includes JDX ORM libraries): Introduction | Download
• ORMCP (for Phase 5): Introduction | Download
• JDBC driver JAR for your database
• Docker (for Phases 3–4)
• pip install rich (optional — nicer terminal output)

Tip

Don't want to install a local JDK/JDX SDK? Docker mode runs the whole tool — Phases 1 and 3 — with only Docker installed.

---

Quick start

Windows

:: 1. Clone the repo to a convenient tools location — do this once, not per project
cd C:\tools
git clone https://github.com/SoftwareTree/orm_skyway_automation.git

:: 2. Create a project directory anywhere you like and enter it
mkdir C:\projects\my_service
cd C:\projects\my_service

:: 3. Copy the template config into your project directory and edit it
copy C:\tools\orm_skyway_automation\orm_skyway_config.json .
notepad orm_skyway_config.json

:: 4. Run Phase 1 (reverse-engineer) + Phase 3 (build Gilhari Docker image) together
python C:\tools\orm_skyway_automation\orm_skyway.py -f orm_skyway_config.json --phase 1+3

:: 5. Start the Gilhari microservice
gilhari\run_docker_app.cmd

:: 6. Verify
curl -s http://localhost:80/gilhari/v1/health/check | python -m json.tool

macOS / Linux

# 1. Clone the repo to a convenient tools location — do this once, not per project
cd ~/tools
git clone https://github.com/SoftwareTree/orm_skyway_automation.git

# 2. Create a project directory anywhere you like and enter it
mkdir ~/projects/my_service
cd ~/projects/my_service

# 3. Copy the template config into your project directory and edit it
cp ~/tools/orm_skyway_automation/orm_skyway_config.json .
nano orm_skyway_config.json   # or your preferred editor

# 4. Run Phase 1 (reverse-engineer) + Phase 3 (build Gilhari Docker image) together
python ~/tools/orm_skyway_automation/orm_skyway.py -f orm_skyway_config.json --phase 1+3

# 5. Start the Gilhari microservice
./gilhari/run_docker_app.sh

# 6. Verify
curl -s http://localhost:80/gilhari/v1/health/check | python -m json.tool

The tool repo (orm_skyway_automation/) and your service project directories are completely independent. You can have as many service projects as you like in different locations — each one just references the same cloned tool.

Want to pause between phases to refine the ORM spec first?

Windows:

python C:\tools\orm_skyway_automation\orm_skyway.py -f orm_skyway_config.json --phase 1

:: (optionally edit config\*.jdx, then recompile with scripts\compile.bat)

python C:\tools\orm_skyway_automation\orm_skyway.py -f orm_skyway_config.json --phase 3

macOS / Linux:

python ~/tools/orm_skyway_automation/orm_skyway.py -f orm_skyway_config.json --phase 1

# (optionally edit config/*.jdx, then recompile with ./scripts/compile.sh)

python ~/tools/orm_skyway_automation/orm_skyway.py -f orm_skyway_config.json --phase 3

---

Connecting an AI agent (ORMCP)

Once your Gilhari microservice is running, connecting it to an AI agent takes just a few minutes.

ORMCP is an MCP server that bridges AI language models to your Gilhari microservice. The agent reasons about your domain objects — Employee, Order, Product — not raw table rows.

pip install ormcp-server

Then add one entry to your AI client config. For Claude Desktop (claude_desktop_config.json):

{
  "mcpServers": {
    "my-ormcp-server": {
      "command": "ormcp-server",
      "args": [],
      "env": {
        "GILHARI_BASE_URL": "http://localhost:80/gilhari/v1/",
        "MCP_SERVER_NAME": "MyORMCPServer"
      }
    }
  }
}

Claude Desktop starts the ORMCP server automatically — no separate terminal needed. You can then ask things like "Show me all orders placed this month" and it queries your database through your curated domain model.

Beta: ORMCP is currently in beta — free to use for testing and evaluation, not yet for production. No account or access request is needed to install it; pip install ormcp-server works directly. See softwaretree.com/products/ormcp for details.

---

Supported Databases

Meaning of "Verified" and "Experimental" Status below

What "Verified" means: the database has been tested against a real instance through ORM_Skyway's core workflow — schema reverse-engineering, business object model generation, REST packaging, and CRUD operations — with results checked directly against the database, not just assumed from a passing test. It does not mean every database version, every JDBC driver, or every JDBC connection property has been exercised — that's true of every entry below, not just newer ones. Known limitations for a specific database, where they exist, are noted individually rather than left implicit.

What "Experimental" means: ORM_Skyway is expected to support the database — through a dedicated connection type or through its GENERIC JDBC fallback — but that support hasn't yet gone through the verification process above.

Database	Status
PostgreSQL	✅ Verified
MySQL	✅ Verified
Oracle	✅ Verified
SQLite	✅ Verified
SQL Server (MSSQL)	✅ Verified
Snowflake	✅ Verified
CockroachDB (PostgreSQL interface)	✅ Verified
SAP HANA	✅ Verified
DB2 (LUW)	✅ Verified
Databricks	⚠️ Experimental
MariaDB	⚠️ Experimental
Spanner (PostgreSQL interface)	⚠️ Experimental
YugabyteDB	⚠️ Experimental

---

Want to go deeper?

• Docker mode — run with only Docker installed
• Configuration file reference
• Phase 1 — Reverse engineering in detail
• Phase 2 — ORM refinement and curation guide
• Phase 3 — Gilhari packaging in detail
• Phase 4 — Testing with curl and Postman
• Phase 5 — ORMCP / AI integration
• Command-line reference
• Project layout
• ORMCP documentation
• Gilhari SDK

---

ORM_Skyway is built on Software Tree's JDX ORM, Gilhari, and ORMCP