Loading ToolBench
Single tool 'health_check' with minimal definition quality. Description is present (58 chars) but lacks actionable context for LLM selection. Input schema is empty object {} with no parameters documented. No output schema documented. Tool does not follow verb_noun naming (should be 'check_health' or similar). No error handling guidance. Code shows inconsistent implementations across mcp_server.py and app/mcp_tools.py, suggesting incomplete integration. This server appears to be a stub/template rather than production-ready.
Check MCP server health. Requires valid API key in x-token header.
No input schema defined. Tool has empty input {} with no parameter documentation.
No output schema documented. LLM cannot infer what fields to expect from response {status, message}. Missing return type documentation.
Description lacks actionable context. 'Check MCP server health. Requires valid API key in x-token header.' does not explain WHEN to call this tool or what the response means. No guidance on error cases (invalid key, server down).
Credentials exposed in tool description. Description mentions 'x-token header' but does not clarify that API key is injected server-side, not passed as parameter. Risk of LLM attempting to pass secrets.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 33 | <=2025-11-25 | v2 |
No error handling guidance. Tool returns {status: error, message: ...} on auth failure but does not guide LLM on recovery steps or retry logic.