Critical failures across all dimensions. The mcp.json file lists only tool names as strings with NO schemas, descriptions, or parameter definitions visible. No descriptions are provided in the source code. The server exposes a public HTTP API but the MCP tool definitions lack the structure required for LLM-safe operation.

• Add complete JSON Schema definitions to mcp.json for each tool, including input parameters with type, description, and constraints (enums, min/max, patterns).
• Write tool descriptions (50 - 200 chars) explaining WHAT each tool does, WHEN to use it, and WHAT it returns. Example: 'generate_song_idea: Generate a personalized children's song concept for a given name and optional theme. Returns title, style, characters, chorus, and lyric outline.'
• Document all input parameters with descriptions. For generate_song_idea, describe 'name' (required, string), 'theme' (optional, string), and 'lang' (optional, enum: en|es|pt).
• Document the output schema with field names, types, and descriptions. Based on bin/cucutime.js, output includes: title (string), concept (string), musicStyle (string), companionCharacter (string), chorusSeed (string), lyricIdeas (array of strings), callToAction (object with label and url), language (string).
• Add error handling guidance. Specify what errors can occur (invalid language, missing name) and what the LLM should do next (e.g., 'If language is unsupported, retry with lang=en').