Single tool 'vibe_check' has critical definition gaps. Input schema is empty object {} with no parameters documented. Description is whimsical but uninformative (23 chars, below 20-char threshold). No parameter descriptions, no output schema documentation, no error handling guidance. Tool relies on undocumented elicitation and sampling capabilities that are not exposed in the schema. This violates core patterns: tool-description, constrained-input, and response-shaper.

• Define explicit input parameters in the schema. If the tool requires a 'vibe_phrase' or similar, declare it with type, description, and constraints (e.g., minLength=3, maxLength=500).
• Rewrite description to be clear and actionable: 'Analyzes the emotional vibe of a phrase and returns a whimsical interpretation. Requires user input via elicitation and LLM sampling. Use when you need to understand or express emotional tone.'
• Document the output schema with typed fields (e.g., vibe_word: string, explanation: string, confidence: number) so the LLM knows what to expect.
• Add parameter descriptions for any inputs (e.g., 'vibe_phrase: The text to analyze for emotional tone. Must be 3-500 characters.').
• Include error recovery guidance: 'If elicitation is unavailable, the session does not support user input. If sampling fails, the LLM provider is unreachable, retry or fall back to a static response.'

Inferred effective spec: <=2025-11-25.

• Relies on Sampling (deprecated) - integrate directly with the LLM provider API