TramAI - governed AI workflows for Java and Kotlin

Deep Dives

This section is for readers who want to understand why TramAI is structured the way it is. These guides go beyond usage and explain module boundaries, orchestration concepts, design constraints, and the direction of the system.

In This Section

GuideFocus
Architecture OverviewSystem shape, the runtime execution flow, and the architectural priorities
Module ArchitectureThe 0.6.0 ten-layer model, the machine-readable catalog and boundary rules, and how the architecture gate enforces them
Module DesignWhich modules to use, how responsibilities are split, and what each layer is for
Design Deep DiveCore design choices and the reasoning behind them
API and Binary CompatibilityWhat the stable freeze covers, what is preview, and how consumer compatibility is proven
Quality GatesThe 0.6.0 gate system — formatting, analysis, safety guards, coverage and mutation ratchets
OrchestrationWorkflow orchestration concepts, step shapes, and execution flow
Orchestration PersistenceCheckpoint, resume, leases, and persistence-related architecture
Orchestrator VisionForward-looking direction for the orchestration model

When To Start Here

Start here when you are contributing to the project, evaluating architecture tradeoffs, or trying to understand how the internals fit together.

If you are evaluating TramAI for an application rather than contributing to it, start with Why TramAI? and Modules Catalog instead.