Working with MCP Integrations

AI agents and LLM-powered tools are becoming part of the developer workflow. But most of them operate on raw files and terminal output — they have no understanding of Unreal project structure, module boundaries, or engine-specific conventions. Gamibase's MCP surface gives these tools structured, project-aware context so they can reason about Unreal work instead of guessing.

MCP (Model Context Protocol) is a machine-facing interface. Where the CLI is designed for human operators and VSCode is designed for editor UX, MCP is designed for tools that call Gamibase programmatically. It exposes the same workflows — project intelligence, diagnosis, review — through structured tool routes that return typed JSON.

• Project structure reads as structured JSON — modules, targets, plugins, dependencies
• Build diagnosis with typed findings, not raw log text
• Symbol search with ownership and hierarchy context
• Config reads with structured key-value output
• Crash and review findings in machine-consumable format

MCP is not a separate product or a simplified API. It runs the same Gamibase runtime as the CLI, with the same trust model, evidence gates, and edition boundaries. A tool calling gamibase via MCP gets the same depth of analysis as an operator running commands in the terminal. The output format is different — JSON instead of text — but the analysis is identical.

{
  "tool": "gamibase_info",
  "result": {
    "project": "MyGame",
    "engine": "5.4",
    "modules": [
      { "name": "MyGameCore", "type": "Runtime", "dependencies": ["Core", "Engine"] },
      { "name": "MyGameEditor", "type": "Editor", "dependencies": ["MyGameCore", "UnrealEd"] }
    ],
    "targets": ["MyGame", "MyGameEditor"]
  }
}

MCP is most valuable when you need to give an external tool real Unreal context. Without it, an AI agent trying to help with a build failure will read raw log text and hallucinate about module structure. With MCP, it gets structured findings, real dependency graphs, and typed project context.

1. AI coding assistants that need project-aware context for Unreal work
2. Custom automation scripts that need structured build or project data
3. Dashboard and reporting tools that consume diagnosis results
4. CI integrations that need typed output for downstream processing