Working repository for a simulation-first RPG engine.
The goal is a headless, turn-based world simulation with bounded local two-dimensional square grids and a regional topology. It can support multiple clients: terminal, web, native UI, automated tests, and AI agents. The frontend is not the source of truth. The simulation core is.
- Turn-based RPG play over local square grids and a regional world topology.
- Simulation depth matters more than graphics.
- Human players and AI agents should use the same action interface.
- Entities perceive only what their senses, knowledge, memory, and position allow them to perceive.
- World changes flow through typed admission or scheduled work, verified atomic authority transitions, domain events, and explicit later causal work.
- Content should be system-driven: materials, body parts, factions, knowledge, environment, tools, rituals, mutations, and social relationships should combine into emergent behavior.
Start with the Documentation Guide. It identifies the role and authority of every documentation layer.
- Vision and Reference Game Vision define product direction and the gameplay pressure the architecture must survive.
- Target Architecture is the only normative cross-system architecture.
- Execution Roadmap maps required capabilities to milestones and validation evidence.
- Rewrite Status records completed milestones and whether a detailed milestone plan is currently active.
- Design Index classifies current, partially superseded, and frozen design inputs.
- Research Index separates accepted dispositions from evidence and open hypotheses.
M1 through M7 build and operationalize the engine foundation. The target roadmap is architecture-complete only after the M8 gameplay-composition suite has demonstrated cross-system depth without adding a second authority path or a speculative universal framework.