Introducing the GameFrame MCP Server: Your Design Docs Inside Your AI Coding Tool

Read Time: 8 minutes

§What Is MCP (and Why Should You Care)?#

Model Context Protocol (MCP) is an open standard that lets AI coding tools talk to external data sources. Think of it as a USB port for your AI assistant: plug in a server, and suddenly the AI can read your project management tool, your database, or in this case, your game design documents.

Without MCP, your AI coding tool only sees your code. It has no idea that the player health is supposed to be 150 (per the GDD you updated last week), that the loot table caps rare drops at 2%, or that the quest system was redesigned in branch "narrative-rework-v2."

With the GameFrame MCP server, your AI assistant can pull that context directly from your vault. No copy-pasting. No stale screenshots. Live data.

§Why This Matters for Game Developers#

Game code does not exist in isolation. Every system you build references design decisions that live in documents:

• —The combat system references stat tables and damage formulas from your GDD
• —The inventory UI references item categories and rarity tiers from your balance sheets
• —The dialogue system references character profiles and narrative arcs

When your AI coding assistant can read those documents alongside your code, it writes better code. It catches inconsistencies ("your code caps health at 100 but the GDD says 150"). It generates implementations that match your actual design, not its best guess.

§Setup in 3 Steps#

Step 1: Create an API Token#

Go to Settings > API Tokens in your GameFrame workspace and create a token with read scope. The token starts with gf_ak_ and is shown only once, so copy it immediately.

Step 2: Add to Your Project#

Create or edit .mcp.json in your project root:

{
  "mcpServers": {
    "gameframe": {
      "command": "npx",
      "args": ["@gameframe/mcp-server"],
      "env": {
        "GAMEFRAME_API_TOKEN": "gf_ak_your_token_here"
      }
    }
  }
}

This works with Claude Code, Cursor, Windsurf, VS Code Copilot, and any MCP-compatible client.

Step 3 (Optional): Set a Default Vault#

If your team works primarily in one vault, set a default so you don't have to specify it on every request:

{
  "mcpServers": {
    "gameframe": {
      "command": "npx",
      "args": ["@gameframe/mcp-server"],
      "env": {
        "GAMEFRAME_API_TOKEN": "gf_ak_...",
        "GAMEFRAME_DEFAULT_VAULT_ID": "your-vault-uuid"
      }
    }
  }
}

§All 10 Tools at a Glance#

The MCP server exposes 10 read-only tools. All data stays read-only; the server never modifies your vault.

The server also exposes a resource at gameframe://vault/{id}/tree that returns the full document tree as structured JSON.

§Example Workflows#

"Implement the stat system based on our GDD"#

Instead of tabbing to the browser, finding the document, and copying values:

1. 1.Your AI tool calls gameframe_search_documents with query "player stats"
2. 2.It reads the content with gameframe_get_document_content
3. 3.It pulls entity data with gameframe_get_document_entities for structured stat definitions
4. 4.It writes the implementation using actual values from your design documents

"What changed in the combat spec since last sprint?"#

1. 1.Call gameframe_get_version_history on the combat document
2. 2.Call gameframe_compare_versions between the version from two weeks ago and now
3. 3.The AI summarizes the design changes and can update your code to match

"Check if my implementation matches the design"#

1. 1.The AI reads your code
2. 2.It calls gameframe_search_entities to find relevant game entities
3. 3.It cross-references your hardcoded values against the source-of-truth design data
4. 4.It flags mismatches: "Your code uses base health of 100, but the GDD specifies 150 as of version 12"

§What's Next#

This is version 0.1 and read-only by design. On the roadmap:

• —Write tools -- Create documents, save new versions, and update change requests from your coding tool
• —Smarter context -- Automatic vault attachment based on your project
• —Prompt templates -- Pre-built prompts for common workflows like "audit my code against the GDD"

Get started: Create an API token in Settings > API Tokens and add three lines of JSON to your project. Your AI assistant will thank you.