feat: Add tool annotations for improved LLM tool understanding

[triepod-ai]

Summary

Adds MCP tool annotations (readOnlyHint, destructiveHint) to all tools to help LLMs better understand tool behavior and make safer decisions about tool execution.

Changes

• Added readOnlyHint: true to get_figma_data tool (fetches data without modifications)
• Added destructiveHint: true to download_figma_images tool (writes files to local disk)
• Added title annotations for human-readable display
• Bump @modelcontextprotocol/sdk from ^1.10.2 to ^1.25.1 for annotation support

Why This Matters

Tool annotations provide semantic metadata that helps LLMs understand tool behavior:

• readOnlyHint: Indicates a tool only reads data without side effects
• destructiveHint: Signals the tool performs write operations (creating/modifying files)
• LLMs can make better decisions about when to use tools and in what order
• Enables safer tool execution by distinguishing read-only from destructive operations

Testing

• Server builds successfully (pnpm build)
• tools/list returns annotations in response
• Annotation values match actual tool behavior:
  • get_figma_data → readOnlyHint: true (only fetches Figma data)
  • download_figma_images → destructiveHint: true (writes files to disk)

Before/After

Before (tools/list response):

{
  "name": "get_figma_data",
  "description": "Get comprehensive Figma file data..."
}

After:

{
  "name": "get_figma_data",
  "description": "Get comprehensive Figma file data...",
  "annotations": {
    "title": "Get Figma Data",
    "readOnlyHint": true
  }
}

🤖 Generated with Claude Code