Loading ToolBench
Give any AI agent live product data through the Apify MCP server. Provides e-commerce scraping capabilities via the Apify E-commerce Scraping Tool over the Model Context Protocol.
Three tools with partial schemas and descriptions. Tool 1 (apify--e-commerce-scraping-tool) has a verbose description (180+ chars) and input schema with 6 parameters, but schema lacks type constraints (no enums for scrapeMode, no min/max for maxProductResults). Tools 2 and 3 (get-actor-run, get-dataset-items) have minimal descriptions (under 100 chars) and sparse parameter documentation. No output schemas documented for any tool. No error handling guidance. Parameter descriptions are generic; none explain expected formats, ranges, or recovery paths. Tool naming uses dashes (non-standard for Python/MCP); verb-noun structure is present but inconsistent (get-actor-run vs apify--e-commerce-scraping-tool). No tool annotations (readOnlyHint, idempotentHint) despite all being read-only operations.
E-commerce Scraping Tool that scrapes product data from e-commerce websites. Accepts input with product URLs or keywords and marketplace specifications, returns normalized product data including name, price, brand, stock status, ratings, and other product attributes.
Poll the status of an Apify Actor run and retrieve updated metadata including run status, dataset ID, and execution statistics.
Fetch items from an Apify dataset by dataset ID, with optional limit on the number of items returned.
No output schemas documented. LLMs cannot plan downstream calls or extract required fields (e.g., product IDs, dataset structure). Violates pattern:tool requirement for documented return types.
Parameter descriptions lack actionable constraints. 'scrapeMode' has no enum; 'maxProductResults' has no min/max bounds; 'waitSecs' has no range. LLMs will pass invalid values without guidance.
No error handling guidance. Tools return errors but do not tell LLM what to do next (retry? lookup? ask user?). Violates pattern:recovery-guide.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 48 | 2026-07-28+ | v2 |
Tool names use dashes (apify--e-commerce-scraping-tool, get-actor-run) instead of underscores. Inconsistent with Python/MCP conventions and harder for LLMs to parse.
No tool annotations despite all tools being read-only. Missing readOnlyHint would help LLMs understand safety and retry semantics.