Loading ToolBench
MCP server for web scraping using the Scrapeless API
Single tool with minimal definition quality. Tool name 'scrape_markdown' is action-verb-based (positive), but description is extremely brief (54 chars, below 10-char minimum for context). Input schema present but lacks parameter descriptions, the 'url' parameter has only a generic description. No output schema documented. No error handling guidance. No parameter constraints (format, length, allowed values). Tool definition appears directly visible in verify_mcp.py, so not inferred.
Scrape a URL and return its content as markdown
Tool description is 54 characters, too brief to guide LLM selection. Should explain WHAT it does, WHEN to use it, and what it returns. Current text 'Scrape a URL and return its content as markdown' lacks context for when to prefer this over other tools or what format the markdown takes.
Input parameter 'url' lacks a description explaining format, constraints, or examples. Should state: 'A valid HTTP/HTTPS URL (e.g., https://example.com). Must be publicly accessible.'
No output schema documented. LLM cannot plan downstream operations or extract fields. Should document: returns object with 'markdown' (string), 'title' (string), 'url' (string), 'status_code' (integer), 'error' (string, optional).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 42 | 2026-07-28+ | v2 |
No error handling guidance. Tool will fail on invalid URLs, timeouts, 403/404 responses, or network errors. Should document: 'Returns error field with reason (invalid_url, timeout, access_denied, not_found, network_error). LLM should retry on timeout or suggest alternative URL on 404.'
No parameter constraints. URL parameter should specify: format (URI), length limits, allowed schemes (http/https only), and whether authentication/API keys are required. Current schema is bare string with no validation hints.