Loading ToolBench
MCP server for the Gomag Public API — AI-powered e-commerce management with full audit logging
25 tools with consistent naming (verb_noun pattern) and complete input schemas. All tools have descriptions (avg 85 chars, within 10-1024 range). However, descriptions are generic and lack LLM-optimized guidance on WHEN to use each tool vs. similar ones. Parameter descriptions are present but minimal (avg 45 chars). No output schemas documented, LLMs cannot predict response structure for chaining. Error handling is absent from tool definitions. Security-sensitive tools (customer_login, customer_change_password) accept plaintext credentials as parameters, violating secret-injection pattern. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk classifications in metadata.
Create one or more product attributes.
List product attributes defined in the store.
Update (patch) existing product attributes.
Return the list of configured shipping carriers/couriers available in the store.
Manually create an AWB record for an order.
Delete an AWB record.
Generate an AWB automatically through the carrier's integrated API.
Plaintext credentials in tool parameters (customer_login, customer_change_password). Passwords passed as parameters will be logged in audit trails and agent traces, violating secret-injection pattern.
No output schemas documented. LLMs cannot predict response structure, making tool chaining difficult. E.g., does customer_list return customer_id or id? Does awb_list return awb_id or awb_number?
Descriptions lack LLM-optimized guidance. Generic descriptions like 'List product attributes' do not explain WHEN to call this vs. similar tools, or what the response structure is. Descriptions should be 50-200 chars with actionable context.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 62 | <=2025-11-25 | v2 |
List AWB (Air Waybill / tracking numbers) records.
Generate a printable shipping label (PDF or similar) for an AWB.
Update the delivery status of an AWB.
Create one or more categories.
Delete categories. Only empty categories (no products, no sub-categories) can be deleted.
List product categories.
Update (patch) existing categories.
Change a customer's password.
Create a new customer account.
Submit a GDPR account-deletion request for a customer. The customer will be flagged for deletion in the store's admin panel. This does NOT immediately delete the account.
List customers registered in the Gomag store.
Authenticate a customer and return session/token data.
Get a list of products that a specific customer has ordered.
Trigger a password recovery e-mail for a customer.
Update an existing customer's details.
Cancel (void) an invoice.
Create (register) an invoice for an order.
Auto-generate an invoice for an order using the store's configured invoicing settings.
No tool annotations despite clear risk classifications. Tools marked DESTRUCTIVE (awb_delete, category_delete, invoice_cancel, customer_delete_request) should have destructiveHint=true. READ_ONLY tools should have readOnlyHint=true. This metadata guides agent safety policies.
No error handling guidance in tool definitions. Tools accept JSON strings (attribute_create, awb_create, invoice_create) but do not document what happens on parse failure, validation error, or API rejection. LLMs need recovery hints.