Four tools with clear, domain-specific purposes (AI-readiness audits, SEO checks, llms.txt validation). All have descriptions (194 - 280 chars, within baseline 34 - 392 range). Input schemas present with typed parameters (string). However, output schemas are NOT documented in the visible code, only return type hints (dict) appear. Parameter descriptions are minimal (e.g., 'The URL of the web page to audit' is generic). No enum constraints, no error handling guidance, no pagination support despite tools that may return large result sets. Tool names follow verb_noun pattern (audit_*, compare_*, check_*) and are unambiguous. Prompts are well-designed (full_site_audit, compare_with_competitor) with clear instructions. Missing: output field documentation, parameter validation rules, error recovery guidance, and per-parameter descriptions beyond the bare minimum.

• Document output schemas for all tools. Specify fields returned (e.g., score: int, checks: array of {id, name, passed, evidence}, fixes: array of {priority, description, effort}). Use TypedDict or Pydantic model annotations visible in docstrings.
• Enhance parameter descriptions with format and constraint details. E.g., 'url (string, required): Full HTTP(S) URL of the page to audit. Must be publicly accessible and return HTTP 200. Timeouts after 30s.'
• Add error handling guidance to tool descriptions. E.g., 'Returns error if URL is unreachable, returns 0-100 score even if some checks fail (partial audit).'
• Implement and document result limits. E.g., 'Returns top 10 prioritized fixes (sorted by impact). Use compare_ai_readiness to see detailed per-check diffs.'
• Clarify tool selection in descriptions. E.g., audit_ai_readiness: 'Use this for a single page's AI-search readiness. Use compare_ai_readiness to benchmark against a competitor.'