Foundgine
Hands-on · first application

Get started with Foundgine

Start with the smallest realistic Supply Chain sample, then drill into the advanced semantic and security proving ground.

What you will build

The starter sample keeps the application small while preserving the important boundaries: API, application, domain, semantics, PostgreSQL and MCP. 

Caller / AI agent
      ↓
MCP / API
      ↓
Application capability
      ↓
Semantic model + authorization
      ↓
Foundgine.Runtime
      ↓
Foundgine.Providers → PostgreSQL

Prerequisites

  • .NET 9 SDK, or JDK 21 + Maven 3.9+ for the Java samples
  • Docker with Compose support
  • A clone of the Foundgine repository

Run the starter — .NET

Use the repository sample as the executable source of truth. The website deliberately avoids copying a long command transcript that can drift from the sample.

  1. Open samples/Foundgine.SupplyChain.
  2. Follow Building the Starter, step by step.
  3. Use Foundgine Supply Chain, Explained when you want the reasoning behind each file and boundary.

Run the starter — Java

The Java port follows the same shape, built as a Maven module instead of an ASP.NET Core project.

  1. Open samples/foundgine-supply-chain.
  2. Follow Building the Java Starter, step by step.
  3. Use Foundgine Supply Chain (Java), Explained for the reasoning behind each file.
  4. For the complete MCP + PostgreSQL path already wired up in Java, continue with samples/foundgine-supply-chain-advanced and its step-by-step tutorial.

What to notice while running it

Semantic boundaryMeaning is separated from persistence and transport.AuthorizationThe host supplies trusted identity and the request is authorized before execution.Execution traceSee the same idea with concrete payloads at every stage.Sample progressionMove from starter to advanced semantics and PenTest coverage.

Go deeper

Once the starter makes sense, move to the advanced semantic sample. Its numbered repository docs cover claims and authorization, high-assurance scenarios, ambiguity/grounding, retrieval strategies and adversarial testing.