Server has 7 tools with basic descriptions and input schemas, but significant quality gaps. Tool names are action-oriented (scrape_, extract_, send_, schedule_, fetch_) which is good. However, descriptions are minimal (10-50 chars for most), parameter descriptions are sparse or missing, output schemas are undocumented, and error handling provides no recovery guidance. No tool annotations (readOnlyHint/destructiveHint). Email tools expose sensitive operations (send_email, schedule_email) without confirmation patterns or permission gates. Web scraping tools lack rate-limit or timeout documentation. Average tool score: 42/100.

• Add tool annotations: mark send_email_tool and schedule_email with destructiveHint=true; mark scrape_url, extract_text_by_query, smart_extract with readOnlyHint=true. This enables agents to reason about side effects.
• Expand tool descriptions to 150-250 chars. Include WHEN to use, prerequisites (e.g., 'Requires Gmail credentials in environment'), and any rate limits or timeouts. Example: 'Scrape a webpage and return markdown content. Use for reading public web pages. Timeout: 30s. Returns up to 100KB of content.'
• Document output schemas for all tools. Define return type (object with fields: success, data, error) and field types. E.g., scrape_url returns {success: bool, content: string, url: string, error?: string}.
• Implement confirmation pattern for send_email_tool and schedule_email. Return a dry-run result with recipient, subject, and body preview; require explicit confirmation before sending. Prevents accidental emails.
• Enhance error messages with recovery guidance. Instead of 'Error scraping URL: timeout', return 'Timeout after 30s. URL may be slow or unreachable. Try a different URL or increase timeout. Retryable: yes.'