Mock Machines

Mock Machines is a Go-based finite-state-machine (FSM) engine. It loads state-machine definitions from YAML and runs agent-based experiments: each entity is an FSM instance with states, probabilistic events, conditions, cross-entity messages, and categorical taxonomies. At runtime entities live in a columnar archetype store (struct-of-arrays) for reflection-free, index-based field access — so it generates millions of rows of believable, perfectly-linked data quickly.

The output is a self-consistent world: every foreign key resolves, every observation is reproducible from the event log, and each scenario ships with a generated REST + MCP interface and an OSI-compatible semantic layer.

Quick Start Validate a scenario and run your first experiment.

Config YAML Reference The anatomy of a scenario definition.

System Architecture How the engine is layered, and how it runs.

Glossary The core vocabulary, defined.

What it is for

• Demo data — generate large, realistic datasets for development and testing of other projects, with referential integrity guaranteed by construction.
• Agent training & evaluation — drive a living, mutating world over MCP or REST so agents operate against state that responds and ticks forward.

It deliberately stops short of a full agent-based scientific simulator: the focus is throughput and coherence, generating thousands-to-millions of rows fast.