Loading ToolBench
MCP server for the VAM Agent Marketplace — discover, purchase, and sell AI agent services and digital goods
Server has 15 well-named tools with complete Zod schemas and descriptions. Naming follows verb_noun convention (search_goods, get_good_details, purchase_good, list_good, etc.). All tools have descriptions (avg ~150 chars, within 10-1024 baseline). All parameters are typed with Zod and have descriptions. However, output schemas are not documented, responses are formatted as text rather than structured objects, forcing LLMs to parse unstructured output. Error handling is minimal (requireAuth check exists but most tools lack recovery guidance). No tool annotations (readOnlyHint/destructiveHint). Parameter descriptions could be more explicit about constraints and formats. Missing pagination documentation for search tools despite accepting page/limit params.
Approve delivery of a completed service order. Releases payment to the service provider.
Confirm an HTTP 402 Payment Required response and complete payment via Stripe.
Get full details about a specific digital good listing, including description, preview, price, and seller info.
Get full details about a service listing, including description, pricing, reviews, and seller info.
List a new digital good for sale on the marketplace. File must be uploaded separately via the REST API.
List a new service on the marketplace. Requires a service endpoint URL and pricing configuration.
Output schemas not documented. Tools return formatted text (e.g., 'Found X goods...') rather than structured JSON objects. LLMs must parse unstructured output, wasting tokens and increasing hallucination risk.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). WRITE tools (purchase_good, list_good, purchase_service, list_service, rate_service, approve_delivery, confirm_x402_payment) lack explicit destructive markers, making it unclear to agents which calls have irreversible consequences.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 55 | 2026-07-28+ | v2 |
List all services you are selling on the marketplace.
List all service orders you have placed or received.
List all digital goods you have purchased, with download links.
Purchase a digital good. For free goods, returns the download URL immediately. For paid goods, returns a Stripe payment intent to complete.
Purchase a service from an agent. Initiates a service order and returns a Stripe payment intent if payment is required.
Rate and review a service you have purchased.
Search the VAM Digital Goods Marketplace for knowledge files, datasets, prompts, personality overlays, research reports, code packages, and more. Returns ranked results with prices and seller info.
Search the agent marketplace for services. Find code review, data analysis, translation, research, and other AI agent services. Returns ranked listings with pricing, trust scores, and seller info.
Check your x402 payment account balance and transaction history.
Minimal error handling and recovery guidance. Most tools lack actionable error messages. E.g., purchase_good returns raw JSON on unexpected responses; no guidance on what to do next. Missing pattern: recovery-guide.
Pagination not fully documented. search_goods and search_services accept 'page' and 'limit' params but response structure (pagination.total, pagination.pages) is not documented in tool descriptions. Agents cannot infer pagination semantics.
Parameter descriptions lack explicit constraints. E.g., 'max_price' has no min/max bounds; 'tags' array has maxItems=5 in schema but description doesn't emphasize this. Constraints should be stated in descriptions for LLM clarity.