AD-Wiki MCP server has 20 tools with consistent naming (verb_noun pattern) and German descriptions. All tools declare input schemas (McpKnowledgeListInputSchema, etc.), but schema details are not visible in source, only schema names are referenced. Descriptions are present (avg ~120 chars) but lack LLM-optimized structure: they do not clearly state WHAT the tool does, WHEN to use it, or what it returns. Parameter descriptions are absent from visible code. Output schemas are undocumented. Error handling and recovery guidance are not evident. Security model (ACL enforcement, token-based access) is mentioned in descriptions but not formalized in tool definitions. Tool composition is sound (single responsibility, clear chains), but parameter constraints and validation rules are not visible.

• Expose full input schemas (parameter types, descriptions, enums, constraints) in tool definitions. Reference schema files or inline JSON Schema objects so LLMs can validate inputs before calling.
• Rewrite tool descriptions to follow the pattern: '[ACTION] [RESOURCE] [WHEN/WHY]. Returns [KEY_FIELDS].' Example: 'Search active standards by keyword. Use this to find policies matching a topic. Returns standard_id, title, and summary.' Keep to 50 - 150 chars.
• Document output schemas for every tool. Specify field names, types, and whether pagination is supported. Include example responses in test files or API docs.
• Add error handling documentation: list possible error codes, classify as retryable/user-fixable/fatal, and provide recovery hints. Example: 'If note_not_found, try search_notes() with a partial title.'
• Implement confirmation or dry-run patterns for write tools. Add an optional 'dry_run' parameter (default false) or require explicit 'confirm=true' for destructive operations.