All 5 tools have minimal descriptions (13-20 chars) and identical single-parameter schemas. Tool names lack action verbs (pizzaz-* instead of show_*, render_*, or display_*). No parameter descriptions provided. No output schemas documented. No error handling guidance. Schemas are present but trivial (single string param, no constraints). This is a template/example server with placeholder implementations, not production-grade tooling.

• Rename tools with action verbs: 'show_pizza_map', 'render_pizza_carousel', 'display_pizza_album', 'list_pizza_options', 'play_pizza_video'. This lets LLMs infer intent from names.
• Expand descriptions to 50-200 chars. Example: 'Display an interactive map of pizza restaurants filtered by topping. Returns map widget URI and restaurant list.' Include when/why to use.
• Add parameter descriptions: 'pizzaTopping (string, required): Pizza topping to filter by (e.g., pepperoni, mushroom, sausage). Case-insensitive.'
• Document output schema with structured fields: { widget_uri: string, restaurant_count: number, restaurants: [{name, address, rating}] }. Enable chaining.
• Add error handling: validate pizzaTopping against allowed values; return 'Invalid topping: got "xyz". Valid options: pepperoni, mushroom, sausage, ...' on failure.