GEMINI.md - Context & Instructions for "Super Mentor" (最强导师)

1. Project Overview

Project Name: Super Mentor (最强导师 / Supervisor Simulator) Type: Simulation / Strategy / Roguelite Game Engine: Godot 4.x (C# / .NET) Platform: Windows (Win32), likely cross-platform capable.

Core Concept: A "survival horror" simulation of academic life. Players manage a lab, recruit students, publish papers, secure funding, and navigate academic politics ("publish or perish").

2. Architecture & Directory Structure

2.1 Game Engine (Godot/C#)

Root: Contains project.godot, .sln, .csproj.
scripts/: The core C# source code.
- Core/: Game systems (TurnSystem, TaskSystem, EconomySystem, SynergySystem) and Interfaces.
- Models/: Data models (StudentModel, TaskModel, etc.) and Definitions.
- [Node classes]: Player.cs, Student.cs (Godot Node behaviors).
resources/definitions/: Data-driven design files.
- .json: Archetypes, Roles, Traits.
- .tres: Discipline specific resources (Godot Text Resource).
scenes/: Godot .tscn scene files.
docs/: Detailed game design documents (GDD).

2.2 Tooling: Numerical Design & Balancing System (NDBS)

Located in tools/ndbs/, this is a web-based tool for designers to edit game data and write scripts.

Backend: Python 3.14 (FastAPI), managed by uv.
Frontend: Vue.js (Vite, Naive UI/Element Plus).
Spec: specs/numerical_design_system.md.

3. Building & Development

3.1 Game (Godot)

Prerequisites: Godot Engine (Mono version), .NET SDK.
Build: Standard Godot build process or via IDE (Visual Studio / JetBrains Rider / VS Code).
Run: Open project.godot in Godot Editor or run via CLI.

3.2 NDBS Tool (`tools/ndbs`)

Backend (tools/ndbs/server):
- Manager: uv
- Run: uv run main.py (or uvicorn main:app --reload)
Frontend (tools/ndbs/client):
- Manager: npm / yarn
- Build: npm run build (Output consumed by Python server).

4. Key Conventions

4.1 C# / Godot

Systems Pattern: Core logic is decoupled from Godot Nodes where possible, residing in scripts/Core implementing IGameSystem.
Rule System: Game logic "rules" (Buffs/Debuffs) are referenced by string IDs (RuleIds) and executed via a RuleManager (using Reflection).
Data Driven: Heavy reliance on JSON/TRES definitions for content.

4.2 Code Style

C#: Standard .NET conventions (PascalCase for public, camelCase for private/locals).
Python: Type hints required (Python 3.14+). Pydantic for data validation.
Frontend: Vue Composition API (<script setup lang="ts">).

5. Current Task Context (As of Jan 2026)

The project is actively implementing the NDBS Tool defined in specs/numerical_design_system.md. Goal: Create a unified service where a Python backend serves a Vue frontend to allow editing of resources/definitions and generation of scripts/Rules/Generated/*.cs files.

Immediate Next Steps:

Initialize the Python backend using uv.
Set up the Vue frontend.
Implement the File System API for JSON/TRES handling.

3.2 KiB Raw Blame History