Server has 7 tools with basic descriptions and some parameter documentation, but significant gaps in schema completeness, parameter descriptions, and error handling guidance. Tool names follow verb_noun convention (good), but descriptions are inconsistent in depth. Most tools lack structured output schemas and error recovery guidance. Parameters like process_id and atom_id have descriptions, but optional parameters (process_id in diagnose_last_execution, limit in list_executions) lack clarity on defaults and constraints. No input validation rules documented. Error handling returns generic exception strings rather than actionable recovery steps.

• Convert all tool responses from free-text strings to structured JSON objects. E.g., list_processes should return {"processes": [{"id": "...", "name": "..."}], "total": N} instead of formatted text. Document the output schema for each tool.
• Add actionable error messages with recovery steps. Replace 'Failed to list processes: {str(e)}' with 'Failed to list processes: {error_type}. {recovery_hint}' (e.g., 'Connection timeout. Verify BOOMI_ACCOUNT_ID is set and Boomi API is reachable.').
• Document parameter constraints explicitly. Add to list_executions: 'limit: integer, 1 - 100 (default 10). Maximum results to return.' Add to diagnose_last_execution: 'process_id: optional string. If empty, diagnoses the most recent execution across all processes.'
• Add confirmation or dry-run support for run_process. Either require an explicit confirm=true parameter, or return a 'pending_confirmation' status that requires a follow-up call to execute.
• Return pagination metadata and chaining IDs. list_processes and list_executions should return total_count and next_cursor (if applicable) so agents can paginate. Include process_id in execution results so agents can call diagnose_last_execution without re-lookup.