WeFact MCP has solid naming conventions (verb_noun pattern) and comprehensive parameter descriptions. All 8 tools have clear, actionable descriptions (avg ~150 chars). Input schemas are present with type definitions and parameter descriptions. However, output schemas are not documented, tools return JSON but the structure is not formally specified. Error handling returns structured JSON with error details but lacks recovery guidance. The generic 'wefact_request' tool is well-designed as an escape hatch but reduces the need for specialized tools. Missing: output schema documentation, error classification (retryable vs fatal), and confirmation patterns for destructive operations (create_invoice, update_debtor, create_debtor).

• Document output schemas for all tools. Example: 'Returns {status: string, debtors: [{DebtorCode: string, CompanyName: string, ...}], totalresults: number}'. This enables LLMs to plan chained calls.
• Add dry-run or confirmation step to create_debtor, update_debtor, create_invoice. Example: add 'confirm=true' parameter that defaults to false; when false, return a preview; when true, execute.
• Enhance error responses with recovery guidance. Instead of '{status: error, errors: [...]}', return '{status: error, errors: [...], recovery: "Try search_debtors() to find the correct DebtorCode", retryable: true}'.
• Formalize invoice_lines schema in create_invoice. Define as array of objects with required fields (ProductCode OR Description+PriceExcl) and optional fields (Number, TaxCode). Use JSON Schema $ref or inline definition.
• Update get_debtor and get_invoice descriptions to state: 'Provide either debtor_code or identifier, not both. If you only have a name, call list_debtors() first.'

Inferred effective spec: <=2025-11-25.

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