Server has 4 tools with complete input schemas and descriptions. Naming follows verb_noun pattern (search_, get_, find_). Descriptions are present but generic (avg 95 chars, below 194-char baseline). Parameters have types and descriptions, but lack actionable constraints. get_article has a critical schema flaw: 'article' is required but neither lawId nor lawNum is required, making the tool unusable without clarification. Output schemas are not documented. Error handling is minimal, no recovery guidance or actionable error messages visible in code.

• Fix get_article schema: change required array to ["article"] and add a note in description: 'Either lawId or lawNum must be provided to identify the law.' Consider using oneOf to enforce mutual exclusivity.
• Document output schemas for all tools. Example for search_laws: 'Returns array of {lawId, lawName, lawNo, promulgationDate}. Max 50 results.' This enables LLMs to chain calls correctly.
• Expand descriptions to 100-150 chars with WHEN guidance. E.g. 'Search laws by keyword when you need to find relevant legislation. Use find_related_laws to discover enforcement orders for a known law. Returns up to 50 matching laws with IDs and dates.'
• Add error recovery hints in tool descriptions. E.g. 'If no results found, try broader keywords or different categories (acts, orders, rules).' This guides LLM retry logic.
• Add format constraints to parameter descriptions. E.g. lawId: 'e-Gov law ID (12-digit alphanumeric, e.g. 503AC0000000035)', article: 'Article number as string (e.g. "2" or "2-1" for sub-articles)'.

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry