Single tool (web_search) with adequate naming and schema but weak description. Tool name 'web_search' follows verb_noun convention and is clear. Input schema is present with proper types (string, integer) and parameter descriptions. However, the tool description (95 chars) is below the 194-char baseline for production tools and lacks critical context: no mention of when to use it vs. alternatives, no guidance on result limits, no error recovery hints. Parameter 'top_n' lacks constraints (min/max bounds). Output schema is not documented, LLMs cannot predict response structure. No error handling guidance. Risk classification (READ_ONLY) is correct but insufficient for production use.

• Expand tool description to 150-250 chars: 'Search the live web for current information using a search engine. Use this when you need up-to-date facts, news, or information not in your training data. Returns up to top_n results with title, URL, and snippet. If no results found, try a simpler query or different keywords.'
• Add numeric constraints to top_n parameter: minimum=1, maximum=20, default=5. Document in schema and description: 'Maximum number of results to return (1-20, default 5).'
• Document output schema explicitly: 'Returns array of objects with fields: title (string), url (string), snippet (string), source (string). Total count and next_cursor for pagination if applicable.'
• Add error handling guidance: 'If search fails due to rate limit, wait 60 seconds before retrying. If no results, simplify the query or ask the user for clarification.'
• Consider adding optional parameters: language (enum: en, es, fr, etc.), region (enum: US, UK, etc.), time_filter (enum: past_day, past_week, past_month) to improve result relevance.