Descriptions are present but minimal (10-50 chars), below the 50-200 char baseline for LLM-optimized descriptions. No parameter descriptions exist (all tools have zero parameters). No output schema documentation. No error handling guidance. Tools are read-only and simple, but lack the rigor expected of production tools.

• Add explicit input schema with type: 'object', properties: {}, and required: [] to each tool definition, even if empty. This signals intentional zero-parameter design rather than omission.
• Expand descriptions to 80-150 characters. Example: 'Get a random Chuck Norris joke from the public API. Use this when the user asks for a funny joke or wants to hear something amusing.'
• Document output schema in tool descriptions or via a separate schema field. Example: 'Returns a text/plain joke string.' or 'Returns {joke: string, source: string}'.
• Add try-catch blocks with actionable error messages. Example: 'Failed to fetch joke from chucknorris.io. The service may be down. Try again in a moment or use get-dad-joke instead.'
• Clarify tool selection in descriptions. Example for get-chuck-categories: 'Lists all available joke categories. Call this first if you want to fetch jokes from a specific category (requires a separate category-specific tool).' This guides LLM composition.