MCP Server for Claude Desktop to Perfex CRM integration
Server has 8 tools with basic schemas and descriptions, but significant gaps limit production readiness. All tools have descriptions (10-100 chars, mostly adequate) and Zod-based input schemas with type definitions. However, parameter descriptions are sparse or missing context. Output schemas are undocumented, responses are JSON stringified but structure is not declared. Error handling returns raw error messages without recovery guidance. No tool annotations (readOnlyHint/destructiveHint). The execute_sql tool exposes dangerous raw SQL without validation or confirmation. Parameter naming lacks consistency (e.g., 'phonenumber' vs 'phone_number'). No pagination support despite list_* tools. Missing enum constraints for status/source IDs.
Create a new client/company in Perfex CRM
Create a new lead in Perfex CRM
Get the schema/structure of a specific database table
Execute a raw SQL query on the Perfex CRM database. Use with caution for modifications.
List clients/companies from Perfex CRM
List leads from Perfex CRM with optional filters
List all database tables in the Perfex CRM database with row counts
execute_sql tool allows raw SQL execution without validation, sanitization, or confirmation. No dry-run mode. Exposes SQL injection risk and enables destructive operations without safeguards.
Output schemas are undocumented. Tools return JSON.stringify(results) but LLM cannot infer field names, types, or structure. No documentation of what list_leads, create_lead, etc. actually return.
Error handling returns raw error.message without recovery guidance. LLM cannot determine if error is retryable, user-fixable, or fatal. No actionable error messages.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 53 | 2026-07-28+ | v2 |
Update an existing lead in Perfex CRM
No pagination support. list_leads and list_clients accept limit but no offset/cursor. Large result sets will blow context window. No total_count or next_cursor returned.
Parameter descriptions lack constraint details. status/source/assigned are numeric IDs but no enum values, valid ranges, or lookup guidance provided. LLM must guess valid IDs.