Loading ToolBench
MCP server for searching and retrieving video game pricing data from PriceCharting, including current prices, sales volume, and recent completed sales across multiple condition grades.
Server has 5 tools with complete input schemas and descriptions, but quality is uneven. All tools are READ_ONLY and well-named with action verbs (search_, get_). Descriptions are present (avg ~120 chars) and explain purpose. However, parameter descriptions are minimal or missing context. Output schemas are not documented, LLMs cannot predict response structure. No error handling guidance. Batch tools lack input validation (max 10 items stated but not enforced). Missing per-parameter type hints in some descriptions. Composition is good: tools chain naturally (search → get_prices/sales). No security issues (read-only, no secrets). Overall: functional but below production baseline.
Get current prices and sales volume for a game, console, or hardware item. Use the slug from search_products.
Get current prices for multiple products at once. Runs up to 3 fetches concurrently. Returns prices keyed by each slug.
Get recent completed sales for a specific condition. Use the slug from search_products.
Search PriceCharting for games, consoles, or hardware by title. Returns name, slug, and current prices.
Search PriceCharting for multiple games/consoles at once. Runs up to 3 searches concurrently. Returns results keyed by each query.
Output schemas not documented. LLMs cannot predict response structure (fields, types, nesting). Forces trial-and-error parsing.
Batch tools (search_products_batch, get_product_prices_batch) state 'max 10' in description but do not validate input. LLMs may pass 50+ items; server behavior undefined.
No error handling guidance. If a slug is invalid or network fails, LLM receives raw error with no recovery hint (e.g., 'Try search_products first').
Parameter 'condition' in get_product_sales has enum constraint but no description explaining what each condition means (loose=used, complete=CIB, etc.).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 54 | 2026-07-28+ | v2 |
Batch tools lack per-item error reporting. If 1 of 10 slugs fails, response structure unclear, does it return partial results or fail entirely?