Critical gaps across all three tools. Tool definitions are inferred from file paths and brief descriptions in a React component (web_sota/src/pages/Tools.tsx), not from explicit MCP server code. No input schemas visible in provided source. Descriptions are extremely brief (10-50 chars), below the 34-char minimum baseline. The tool_dispatch.py snippet shows tool registration infrastructure but no actual tool implementations or parameter schemas. FastMCP framework is used but tool definitions are not shown in the source code provided.

• Provide explicit tool registration code showing @mcp.tool() decorators with full input schemas (JSON Schema with types, descriptions, enums, constraints) for each parameter. Example: @mcp.tool() async def separate_stems(audio_file_path: str, model: str = 'htdemucs') -> dict.
• Expand tool descriptions to 50-200 characters following LLM-optimized format: 'WHAT it does, WHEN to use it, WHAT it returns.' Example: 'Separate audio into stems (vocals, drums, bass, other) using AI. Use when you need isolated tracks. Returns paths to stem files.'
• Rename tools with clear action verbs: 'separate_stems' (not 'mixx_stems'), 'update_skin' or 'apply_skin' (not 'mixx_skin'), 'control_deck' or 'send_deck_command' (not 'mixx_deck').
• Document output schemas for each tool. Example: mixx_stems returns {stems: {vocals: string, drums: string, bass: string, other: string}, duration_seconds: number}.
• Add parameter descriptions and constraints. Example: audio_file_path (string, required, format: path to .mp3/.wav/.flac file), model (enum: 'htdemucs'|'mdx'|'mdx_extra', default: 'htdemucs').