24 tools with consistent naming (verb_noun pattern: list_, get_, activate_, etc.) and present descriptions. However, descriptions are minimal (10-50 chars, well below the 194-char baseline), lack context on WHEN to use each tool, and omit prerequisites. Parameter schemas are properly typed with Zod but descriptions are sparse. Output schemas are undocumented, LLMs cannot predict response structure. Error handling is absent; piaFetch returns raw API responses without recovery guidance. No tool annotations (readOnlyHint/destructiveHint) despite clear risk levels. Composition is sound (single responsibility per tool), but parameter descriptions fail to explain constraints, formats, or dependencies.

• Document output schemas for all tools. Specify field names, types, and whether results are paginated. E.g., pia_list_agents returns {value: Agent[], totalCount: number, nextLink?: string}.
• Expand descriptions to 100-200 chars. Add WHEN to use each tool and any prerequisites. E.g., 'List all PIA agents. Call this first to discover available agents before activating or uninstalling. Supports pagination via top/skip.'
• Add tool annotations: mark pia_uninstall_agent, pia_activate_agent, pia_deactivate_agent, pia_trigger_automation, pia_export_repository, pia_import_repository with destructiveHint=true or idempotentHint=true.
• Enhance error handling in piaFetch. Return structured errors with actionable guidance: {error: true, code: 'NOT_FOUND', message: 'Agent 123 not found. Call pia_list_agents to see available agents.', retryable: false}.
• Document parameter constraints in descriptions. E.g., 'agentId (number, 1-999999): The numeric ID of the agent. Call pia_list_agents to discover valid IDs.' For variableInputs, specify the expected object structure.