MCP server for industry AI use cases including sales and HR operations
Four tools with basic schemas and descriptions, but significant gaps in parameter documentation, output schema clarity, and error handling. Tool names follow verb_noun convention (good), but descriptions are minimal (avg 60 chars, well below 194-char baseline). No parameter descriptions for most inputs. No documented output schemas. No error guidance. No input validation hints. Schemas are present but sparse, parameters lack type info beyond 'string'. No tool annotations (readOnlyHint/destructiveHint). Two WRITE tools lack confirmation/dry-run patterns.
Adds a new absence to the HR System
Adds a visit report to Salesforce
Gets the user current contingent of a leave type e.g. holiday
Automatically processes an order entry and saves it to the ERP-System
Parameter descriptions missing or trivial. 'visit_report' param has only 'The visit report which should to be added' (no format, length, or structure guidance). LLMs cannot infer expected input format.
No output schemas documented. Tools return plain text results (e.g. 'The absence ... is successfully added'). LLMs cannot parse structured data or chain results to downstream tools.
Destructive tools (add_visit_report_to_salesforce, process_automatic_order_entry, add_absence_to_hr) lack confirmation/dry-run patterns. No tool annotations (destructiveHint). Agents cannot safely preview changes before committing.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 41 | <=2025-11-25 | v2 |
Tool descriptions are 30 - 50 characters (well below 194-char baseline). 'Adds a visit report to Salesforce' lacks context on when to use it, what format is expected, or what happens on success/failure.
No error handling or recovery guidance. Handlers return hardcoded success messages (e.g. 'successfully added') with no validation, no error classification, no actionable failure messages.