Simple MCP server demonstrating tools and resources
Two tools with minimal documentation. 'add' has a basic description (18 chars, below 20-char threshold) and proper schema. 'query_data' has adequate description (31 chars) but critical security issues: accepts raw SQL without sanitization, no input validation, no error guidance beyond generic exception handling. Neither tool has output schemas documented. No parameter constraints (enums, ranges). Missing error recovery guidance. No tool annotations (readOnlyHint/destructiveHint). Composition is reasonable (two separate concerns) but security posture is poor for a data-access tool.
Add two numbers
Execute SQL queries safely
query_data accepts raw SQL without sanitization or validation. No input constraints, no prepared statements, no SQL injection protection. LLM can pass arbitrary SQL including DROP TABLE.
No output schemas documented for either tool. LLMs cannot plan downstream operations or extract structured data. 'add' returns int, 'query_data' returns string, both undocumented.
query_data description (31 chars) claims 'safely' but provides zero safety mechanisms. No rate limiting, no query timeout, no permission checks. Misleading claim.
add description is 18 characters ('Add two numbers'), below the 20-char minimum. Lacks context on when to use vs other arithmetic tools, prerequisites, or return semantics.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 51 | 2026-07-28+ | v2 |
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). 'add' is read-only and idempotent; 'query_data' is destructive (WRITE risk). LLM cannot infer safety properties.