Single tool with basic schema and minimal descriptions. Tool name 'crawl_urls' is verb-based and clear, but the description (58 chars) lacks context on when to use it vs alternatives, prerequisites, or return structure. Input schema is present with proper types but parameter description is generic (23 chars). No output schema documented. No error handling guidance. No parameter constraints (e.g., URL format validation, array size limits). Lacks idempotency guarantees and state-modification clarity.

• Add output schema to tool definition: document that results are markdown strings, include field names (url, markdown_content, error_status), and explain citation format.
• Expand tool description to 100-150 chars: 'Crawl one or more URLs and extract markdown content with inline citations. Use for web research, content extraction, and link analysis. Returns markdown_v2 format with source citations.'
• Add constraints to 'urls' parameter: type array, minItems 1, maxItems 10, items pattern '^https?://', description 'Array of valid HTTP(S) URLs to crawl (1-10 URLs per request)'.
• Implement structured error responses: categorize as 'timeout' (retryable), 'invalid_url' (user-fixable), 'service_error' (fatal), and include recovery hints like 'Try with fewer URLs or check URL format.'
• Document idempotency: add note 'This tool is idempotent, repeated calls with the same URLs return consistent results. Safe to retry on transient failures.'