MCP Scan tool - Security scanner for Model Context Protocol servers and tools
This MCP server has fundamental quality issues across naming, descriptions, schemas, and error handling. Of the 9 tools present, most lack proper parameter documentation and error guidance. Tool names are generic (add, subtract, multiply, divide) and lack action verbs that would help LLMs understand intent. The 'store_value' tool has a vague, misleading description ('Store here all important values!') with an empty input schema, suggesting incomplete implementation. Parameter descriptions exist but are minimal (e.g., 'first number to add'). No tool returns documented output schemas, limiting downstream composition. Error handling is absent, the divide tool mentions 'Cannot divide by zero' but lacks structured error classification or recovery guidance. The server uses fastmcp framework with STDIO transport only, capping maximum potential regardless of definition quality.
Add two numbers.
Divide two numbers.
Calculate the greatest common divisor (GCD) of two integers.
Return True if n is a prime number, False otherwise. Args: n: integer to check for primality
Calculate the least common multiple (LCM) of two integers.
Multiply two numbers.
Store here all important values!
store_value has an empty input schema and a vague, non-actionable description ('Store here all important values!'). The description does not explain what the tool actually does, when to use it, or what it returns. This violates the requirement that every parameter must have a description and every tool must explain its purpose.
Generic tool names (add, subtract, multiply, divide) lack action verbs that convey intent. While 'add' technically starts with a verb, the names are too generic and do not follow the recommended verb_noun convention (e.g., calculate_sum, compute_product). This makes it harder for LLMs to disambiguate similar tools and understand their purpose at a glance.
Parameter descriptions are minimal and lack context. For example, 'first number to add' and 'second number to add' do not explain the expected format, range, or any constraints. The rubric specifies that descriptions should include 'expected format, range, and allowed values' (e.g., 'Integer >= 0, <= 1,000,000'). Current descriptions are under 30 characters and cannot guide LLM behavior.
Inferred effective spec: 2026-07-28+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 44 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 45 | - | v1 |
Subtract two numbers.
Get current weather for a location.
No output schemas are documented for any tool. The rubric requires: 'Document the output schema. LLMs need to know what fields to expect so they can plan downstream tool calls and extract the right data.' Without this, agents cannot reliably compose tools together or validate responses.
Error handling is absent or incomplete. The divide tool mentions 'Cannot divide by zero' in the source code, but there is no error classification, recovery guidance, or structured error response format. The rubric requires: 'Error responses must tell the LLM what to do next' and 'Categorize errors as retryable, user-fixable, or fatal.' Currently, no tool provides this.
Tool descriptions are 10-60 characters and lack completeness. The rubric specifies: 'Keep descriptions between 10 - 1024 characters' but also: 'Write descriptions as if prompt-engineering. State WHAT the tool does, WHEN to use it, and any prerequisites.' Current descriptions (e.g., 'Add two numbers.', 'Get current weather for a location.') are too terse to guide LLM selection or explain dependencies.
No input validation rules documented. For example, the divide tool accepts any integers but does not specify: what happens if b=0 (beyond a code comment), whether negative numbers are allowed, or whether there are min/max bounds. Numeric parameters should declare 'Specify minimum and maximum for numeric parameters (e.g. page_size 1 - 100, days 1 - 365).'