SEO Memory MCP has 5 tools with complete input schemas and descriptions. Naming follows verb_noun convention (seo_memory_query, seo_skill_route, etc.). Descriptions are present (avg ~120 chars) and explain purpose. However, parameter descriptions are minimal or missing context about expected formats and constraints. Output schemas are not documented, LLMs cannot predict response structure. No error handling guidance. Tool purposes are domain-specific but lack actionable context for when to use each vs. alternatives. Composition is reasonable (each tool has one responsibility), but chaining requirements are undocumented.

• Document output schemas for all 5 tools. Include field names, types, and descriptions. Example: seo_memory_query returns {results: [{score: number, path: string, heading: string, skill: string, source: string}], total_count: number}.
• Convert 'role' and 'skill' parameters to enums with explicit valid values (e.g., role: enum[pinchy, coral, kelp]) or add format constraints to descriptions.
• Add error handling examples: 'If no results found, try broadening the query or check seo_snapshot_info to see available skills.' Include HTTP status codes and recovery guidance.
• Clarify tool composition: document when to use seo_skill_route vs seo_task_pack vs seo_execution_plan. Add a 'See also' section to each tool description.
• Add pagination support to seo_memory_query if result sets can exceed 20 items. Include offset/limit parameters and total_count in response.