Server has 22 tools with consistent naming (verb_noun pattern) and complete input schemas. All tools have descriptions and parameters are typed with enums where appropriate. However, descriptions are generic and lack LLM-optimization guidance (e.g., 'List all companies' vs 'Retrieve paginated list of companies; use filters to narrow results by name or date'). Output schemas are not documented, LLMs cannot predict response structure. Error handling is minimal (generic error messages without recovery guidance). Missing tool annotations (readOnlyHint, destructiveHint). No documentation of which tools are idempotent or have side effects.

• Add output schema documentation to every tool. Example: 'Returns: {companies: [{id: string, name: string, created_at: ISO8601, updated_at: ISO8601}], total: number, page: number}'
• Expand tool descriptions to 100-200 chars with WHEN guidance. E.g., 'List all companies with optional filtering by name or date. Use this to discover company IDs before calling get_company or create_ticket. Supports pagination (page, pageSize 1-100).'
• Add tool annotations via mcp-go API: mark create_* tools with destructiveHint=true, list_* with readOnlyHint=true, and idempotent tools with idempotentHint=true.
• Enhance error messages with recovery guidance. E.g., 'Failed to list companies: invalid filter syntax. Use simple key-value pairs (name: "Acme") or comparison operators ($gt, $lt). See docs for examples.'
• Document filter syntax with examples in parameter descriptions. E.g., 'Filter by status: {"status": "open"} or {"status": {"$in": ["open", "pending"]}}. See list_tickets for full operator reference.'