16 tools with mixed quality. Naming is verb-forward and clear (get_, create_, update_, check_, track_, validate_). Descriptions present but often generic (avg ~80 chars, below 194-char baseline). Schemas visible with proper types and enums, but many parameters lack descriptions. Output schemas not documented. Error handling minimal, no recovery guidance or categorization. Tool composition is reasonable (single responsibility mostly observed), but parameter descriptions are sparse, forcing LLMs to infer intent.

• Add output schema documentation to all GET/list tools. Example: 'Returns: {character_id: int, name: string, character_type: enum, ...}' so LLMs know what fields to extract.
• Expand parameter descriptions to 50-100 chars each. For enums, explain the semantic difference: 'knowledge_state: "knows" = character has confirmed info; "suspects" = has partial/uncertain info; "unaware" = no knowledge.'
• Implement pagination for list tools: add 'limit' (default 20, max 100) and 'offset' (default 0) params. Document in description: 'Returns up to 20 results; use offset to fetch more.'
• Add error recovery guidance. Example: 'If character not found, call get_characters(series_id) to list available characters and verify the ID.'
• Document idempotency and side effects. Mark which tools are read-only vs. write (already done via Risk field), and clarify if repeated calls with same params produce same result.