Skip to content
GenLabZ Docs

What is GenLabZ?

GenLabZ is a platform for building hands-on computing projects that students complete in a managed AWS sandbox. As an author, you define the project — the objectives, the steps, the solution code, and the evaluation logic — and the platform handles delivery, testing, and student guidance.

The authoring model

You work with three artefacts:

  • A structured spec (spec.yaml) that defines your project’s objectives, steps, handle wiring, and metadata
  • Solution code that implements each step — the reference implementation the platform uses to provision resources
  • Evaluation code that verifies student work — checking that the right resources exist and are correctly configured

Once published, the platform runs your solution code to set up the environment, presents the project to students, evaluates their work against your evaluation code, and generates hints when they get stuck.

Three-layer architecture

The authoring toolchain has three layers:

1. Domain model

The foundation. This is the spec schema, the code contracts for solution and evaluation functions, and the handle wiring system that connects outputs from one step to inputs of later steps. Everything you author lives here — it’s IDE-agnostic and version-controlled.

2. MCP server

The GenLabZ Author MCP server exposes the authoring workflow as tools callable from any MCP-compatible IDE. It handles project initialisation, spec validation, step-level testing, and publishing. You don’t call it directly — your IDE does, on your behalf.

3. IDE-specific workflow

The AI-assisted authoring experience layered on top of the MCP server. The Kiro integration, for example, provides a guided six-phase workflow — from intake through to testing — using the MCP tools under the hood. Other IDEs can connect to the same MCP server with their own workflow.

What you can do as an author

  • Define projects — structure a project as objectives and steps, declare resource handles, and write the spec
  • Author with AI assistance — use the Kiro power (or another MCP-compatible IDE) to generate objectives, steps, instructions, and code with guided prompts
  • Validate deterministically — run validate_project and validate_step at any point to catch schema errors, handle mismatches, and contract violations before testing
  • Test automatically — run solution, evaluation, and cleanup code against a real AWS sandbox to confirm everything works end to end
  • Publish — push a validated project to the registry, where students can access it

Next steps

  • Set up your IDE — install the GenLabZ Author power in Kiro and configure the MCP server connection
  • Quickstart walkthrough — create your first project from scratch, following the six authoring phases