Barcode / GTIN validation MCP (EAN-13, UPC-A, EAN-8, GTIN-14).
Two well-defined tools with clear, specific descriptions (194 - 250 chars each). Both have proper input schemas with typed parameters and descriptions. Tool names follow verb_noun convention (validate_gtin, gtin_check_digit). Error handling returns structured responses with reason fields. Main gaps: no output schema documentation, no parameter constraints (enums/ranges), no error recovery guidance, and descriptions lack WHEN/WHY context for LLM selection.
Compute the mod-10 check digit for GTIN/EAN/UPC data digits (all digits EXCEPT the check digit). Use to generate or repair a barcode.
Validate a barcode number — EAN-13, UPC-A (12), EAN-8, or GTIN-14 (keyless, offline). Checks the mod-10 check digit, reports the format, normalizes to GTIN-14, and returns the GS1 prefix + issuing country/region. Spaces/dashes ignored. Validates the number, not the product.
Output schema not documented. LLMs cannot predict response structure (fields like valid, format, gtin14, gs1_prefix, issuing_region). Undocumented outputs force LLMs to guess field names for downstream reasoning.
Parameter descriptions lack format/constraint details. 'code' param says 'e.g. "0036000291452"' but does not state: accepts 8/12/13/14 digits, spaces/dashes stripped, must be numeric. LLMs may pass invalid formats.
Descriptions lack WHEN/WHY context. 'Validate a barcode' does not explain: when should LLM call this vs gtin_check_digit? What is the use case? This forces LLMs to infer intent from name alone.
Error handling returns structured responses but lacks recovery guidance. When check digit fails, response says 'expected X' but does not suggest 'call gtin_check_digit to repair' or next steps.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 69 | 2026-07-28+ | v2 |
No tool annotations (readOnlyHint, idempotentHint). Both tools are read-only and idempotent, but this is not declared in schema. Agents cannot infer safety properties.