Server has 7 tools with complete input schemas and descriptions. Naming follows verb_noun convention (get_, initiate_, generate_, execute_). Descriptions are present (avg ~120 chars) but lack depth on WHEN to use tools and prerequisites. Parameters are typed and mostly described, but lack constraints (enums, ranges, formats). Output schemas are not documented, LLMs cannot plan downstream calls. Error handling returns text but lacks recovery guidance or categorization. No tool annotations (readOnlyHint/destructiveHint). Missing pagination for list-like operations.

• Document output schema for each tool. Example for initiate_job: 'Returns {jobExecutionId: string, createdAt: ISO8601, status: string}'. This enables LLMs to chain calls without guessing.
• Add tool annotations to CallToolRequestSchema response. Mark initiate_job, generate_file_upload_url, execute_job with destructiveHint=true; mark get_* tools with readOnlyHint=true. This signals intent without requiring description parsing.
• Enhance error responses with recovery guidance. Instead of 'Error: Workflow not found', return 'Workflow not found. Call list_workflows() to see available workflows, or verify the workflowId parameter.'
• Add constraints to parameter descriptions. E.g., fileExtension: 'File extension with dot (e.g., .pdf). Must be one of: .pdf, .jpeg, .png, .docx, .csv, .xlsx, .txt, .json, .html, .xml'.
• Add dependency hints to descriptions. E.g., execute_job: 'Requires jobExecutionId from initiate_job. Use jobPayloadSchema from get_workflow_details to structure jobPayloadSchemaInstance correctly.'