feat: initialize tfcode project structure

- Add README.md as living documentation
- Add tfcode.json schema and config template
- Add FORK_MANAGEMENT.md with mirror-based fork strategy
- Add scripts/rebrand.sh for reapplying branding after upstream merges
- Add packages/tf-sync Python module using official ToothFairyAI SDK
- Add packages/tf-mcp-bridge TypeScript module (stub)
- Multi-region support (AU, EU, US)
- Tool sync: MCP servers, Agent Skills, Database Scripts, API Functions

packages/tf-sync/src/tf_sync/mcp.py

@@ -0,0 +1,54 @@
+"""
+MCP server sync module for tfcode.
+Uses the official ToothFairyAI Python SDK.
+"""
+
+from pydantic import BaseModel
+
+from tf_sync.config import TFConfig
+from tf_sync.tools import SyncedTool, sync_tools, ToolType
+
+
+class MCPServerSyncResult(BaseModel):
+    """Result of MCP server sync operation."""
+    
+    success: bool
+    servers: list[SyncedTool] = []
+    error: str | None = None
+
+
+def sync_mcp_servers(config: TFConfig) -> MCPServerSyncResult:
+    """
+    Sync MCP servers from ToothFairyAI workspace.
+    
+    MCP servers are tools with isMCPServer=true.
+    Credentials stay in TF and are accessed via tf_proxy.
+    
+    Args:
+        config: TFConfig instance
+    
+    Returns:
+        MCPServerSyncResult with synced MCP servers
+    """
+    result = sync_tools_by_type(config, [ToolType.MCP_SERVER])
+    
+    if not result.success:
+        return MCPServerSyncResult(
+            success=False,
+            error=result.error,
+        )
+    
+    # Get MCP servers from tools with isMCPServer
+    mcp_servers = [
+        t for t in result.tools 
+        if t.is_mcp_server
+    ]
+    
+    return MCPServerSyncResult(
+        success=True,
+        servers=mcp_servers,
+    )
+
+
+# Import from tools module
+from tf_sync.tools import sync_tools_by_type