Agent that finds order status for a material used in the building renovation for a home owner, powered by Google ADK and MCP protocol
This MCP server has a single tool with a legitimate input schema, but exhibits critical quality gaps. The tool has an adequate description (78 chars), proper parameter typing with description, and a clear semantic purpose. However, the implementation itself is problematic: the tool is loaded from an external ToolboxSyncClient rather than implemented within the MCP server itself, making the server a thin wrapper that cannot be audited for proper MCP compliance. No output schema is documented, error handling is absent, and the tool definition is inferred from a dynamic load pattern rather than explicit registration. The server also lacks any prompts, resources, or structured capability declarations. For a production MCP server, this represents a low-confidence integration.
Get the status of order of a specific material or item by extracting the name of the material from user input text for building renovation order tracking
Tool implementation delegated to external ToolboxSyncClient; tool definition cannot be audited within MCP server source code
No output schema documented; LLM cannot infer return structure or plan downstream chaining
No error handling or recovery guidance; hard failures will not guide agent toward corrective action
Parameter 'material_name' accepts unbounded free-form strings; should enforce constraints (length, character set, or valid material enums) to reduce hallucination
No tool annotations (readOnlyHint, idempotentHint, destructiveHint) declared; MCP clients cannot optimize caching or retry logic
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 47 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 11 | - | v1 |