Two tools with minimal definition quality. 'ask' has a basic description and single string parameter with minimal context. 'cleardm' lacks any description entirely and has no parameters. Neither tool documents output schemas, error handling, or recovery guidance. Parameter descriptions are trivial (e.g., 'What would you like to ask?'). No enum constraints, no pagination support, no structured output documentation. Tool names lack clarity on what data is returned or how to chain results. Missing all 54 pattern guidance on descriptions, schemas, and error recovery.

• Add a description to cleardm: 'Clear all bot messages from a direct message channel. This is a destructive operation that cannot be undone. Only works in DMs.'
• Document ask output schema: specify what fields the Mastra response contains (e.g., text, metadata, sources) so LLMs know what to extract.
• Enhance ask parameter description: 'A question about Mastra (max 2000 characters). Returns a text response from the Mastra AI system.'
• Add error handling guidance: instead of 'Sorry, I encountered an error...', return structured errors like 'Failed to generate response: [reason]. Try rephrasing your question or contact support.'
• Add tool annotations: mark cleardm with destructiveHint=true and ask with readOnlyHint=true so LLMs understand safety implications.