Server has 6 tools with consistent verb-noun naming (search_*, get_*) and basic schemas via Zod. All tools are READ_ONLY and descriptions are present but brief (20-60 chars). Input schemas are properly defined with required fields and descriptions. However, output schemas are not documented, LLMs cannot predict response structure. Parameter descriptions lack detail on format/constraints. No error handling guidance or recovery paths documented. Composition is sound (single responsibility per tool), but descriptions fall short of LLM-optimization baseline (50-200 chars per pattern:tool-description).

• Expand all tool descriptions to 80-150 chars. Include WHAT the tool does, WHEN to use it vs similar tools, and what it returns. Example: 'Search for atoms (entities) in the Intuition knowledge graph by label or value. Use this to discover entities before retrieving detailed account info. Returns matching atoms with IDs and labels.'
• Document output schemas for all tools. Add a 'Returns' section to each tool description listing key fields (e.g., 'Returns: {atoms: [{id, label, value}], total_count}').
• Add format/constraint guidance to search parameters. Example: 'likeStr: Search string (1-100 chars, case-insensitive). Matches against atom labels and values. Partial matches supported.'
• Add error handling guidance. Example: 'If no atoms found, try a shorter or more general search term. Use search_account_ids to find accounts instead.'
• Document pagination for search tools. Specify if results are capped, and if so, at what limit. Add offset/limit parameters if the API supports them.

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

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