Loading ToolBench
CleanPage — a paid MCP tool that turns any URL into clean, structured, agent-optimized content. Pay-per-call via x402 (USDC on Base).
CleanPage demonstrates solid definition quality with complete schemas, clear descriptions, and thoughtful parameter design. The primary tool (clean_page) has well-documented inputs with proper constraints (max_length capped at 200k, timeout_ms at 20s). Three free discovery tools (health, list_capabilities, pricing) are well-named and serve clear purposes. However, output schemas are not documented, the tool description mentions what is returned (markdown, plain text, metadata, links, images, facts array, quality score) but no formal JSON schema is provided for the response structure. This forces LLMs to infer output shape. Additionally, error handling guidance is minimal, the description mentions 'clear typed errors' but does not explain what those errors are or how to recover from them. The paid tool model is innovative but adds complexity that could be better explained in parameter descriptions.
Fetches a URL and returns clean, structured, agent-optimized content: readable markdown, plain text, metadata (author/date/site/word count), extracted links and images, an optional structured facts array, and a 0-100 quality score. Handles common failure modes (paywalls, bot-walls, empty JS shells) with clear typed errors instead of silently returning junk. PAID TOOL: $0.02 USDC per successful call via x402 (USDC on base-sepolia). Call with no payment first to receive the exact payment requirements (HTTP 402), then retry with an X-PAYMENT header. Failed extractions are not charged.
Free. Returns service status, version, and uptime signal. Use to check the service is reachable before paying for anything.
Free. Returns the full list of tools this server offers, including which are paid, their prices, and their input schemas — so an agent can decide what to call before spending anything.
Free. Returns current pricing, accepted payment network/asset, and the x402 payment flow details for paid tools.
Output schemas not documented. Tool descriptions mention returned fields (markdown, plain text, metadata, links, images, facts, quality score) but no formal JSON schema is provided. LLMs cannot plan downstream operations or validate response structure.
Error handling guidance missing. Description states 'clear typed errors' but does not enumerate error types, conditions, or recovery steps. LLMs cannot self-correct on failure.
Payment flow complexity underexplained in tool parameters. The x402 payment model (call without payment → receive 402 → retry with X-PAYMENT header) is mentioned in description but not formalized as parameter constraints or error codes. LLMs may not understand the retry pattern.
Discovery tools (health, list_capabilities, pricing) lack output schema documentation. Agents cannot know what fields to expect or how to parse responses.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | B | 79 | 2026-07-28+ | v2 |
Parameter 'options' is optional but contains optional sub-fields with no clear defaults. LLMs may not understand which combinations are valid or what happens when omitted.