Two tools present with basic fastmcp registration. Both have descriptions and input schemas visible in operations.py. However, descriptions are verbose and lack clarity on WHEN to use each tool. Parameter descriptions are present but lack format constraints and validation guidance. No output schema documentation. Error handling returns generic dicts without recovery guidance. Tool naming follows verb_noun pattern (list_skus, update_sku_qty) but descriptions contain instructional text better suited to agent prompts, not tool docs. Average tool score: 52.

• Document output schemas explicitly. For list_skus, specify: {'skus': [{'id': string, 'name': string, 'quantity': integer, 'cost': number}, ...]}. For update_sku_qty, specify: {'success': boolean, 'sku_id': string, 'new_quantity': integer} on success; {'error': string, 'suggestion': string} on failure.
• Shorten and clarify tool descriptions. Remove instructional text ('return only consumer friendly information'); instead, add it to system prompts or agent instructions. Tool description should be: 'Retrieves all SKU records from inventory. Use this to answer questions about available items, pricing, and quantities. Accepts optional sku_name filter (default: wildcard returns all).'
• Add recovery guidance to error responses. Instead of {'error': 'SKU data file not found.'}, return {'error': 'SKU data file not found.', 'suggestion': 'This may indicate a configuration issue. Verify CSV_FILE_PATH is set correctly and the file exists at the expected location.'}
• Add tool annotations (destructiveHint, idempotentHint, readOnlyHint) to the fastmcp registration. Decorate update_sku_qty with @mcp.tool(destructiveHint=True, idempotentHint=False) to inform agents that calls are irreversible and must not be retried blindly.
• Add pagination to list_skus. Accept optional parameters: limit (default 50, max 500), offset (default 0). Return {'skus': [...], 'total_count': integer, 'offset': integer, 'limit': integer}. Document in description: 'Returns up to 50 SKUs per call; use offset/limit for large datasets.'

Inferred effective spec: 2026-07-28+.

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