Loading ToolBench
MCP server exposing creative services, packages, pricing, and contact information for DY Creative (大瑀创意科技)
Both tools have minimal definitions with empty input schemas and generic descriptions. Tool names follow verb_noun convention (get_*), but descriptions lack actionable context for LLM selection. No parameter documentation, no output schema specification, and no error handling guidance. The drift-check script itself is well-engineered for CI validation, but the MCP tool definitions exposed to clients are severely underdeveloped. This server would not pass production code review.
Retrieve contact information including phone, email, and address
Retrieve available service packages with pricing information, including content packages and AI vision packages
Empty input schemas ({}). Both tools declare no parameters, but LLMs cannot infer what data they return or how to chain results downstream.
Generic descriptions under 50 characters. 'Retrieve available service packages with pricing information...' and 'Retrieve contact information...' lack context for when/why to call these tools vs alternatives. No mention of return structure, prerequisites, or use cases.
No output schema documented. LLMs cannot plan downstream tool calls or extract required fields (e.g., package names, prices, contact fields). Responses are opaque to the agent.
No error handling or recovery guidance. If get_service_packages fails or returns unexpected structure, the LLM has no actionable next step. No categorization of retryable vs fatal errors.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 42 | 2026-07-28+ | v2 |
Tool definitions inferred from drift-check script, not from explicit MCP server registration code. No visible server.py, handler registration, or schema definitions in the provided source. Actual tool implementation is opaque.