A C implementation of the Model Context Protocol (MCP) with server and client libraries, supporting stdio and HTTP transports
cMCP exhibits severe definition quality issues. Tool definitions are duplicated (echo and add appear twice with inconsistent descriptions), lack parameter descriptions entirely, and have minimal or trivial descriptions (e.g., 'echo', 'fixed sleep'). No input schemas are visible in the provided source code, only inferred from the tool list. The 'sleep' tool has no schema shown. Descriptions range from 5-80 characters, well below the 194-char baseline for production tools. No output schemas documented. No error handling guidance. This is a reference/benchmark implementation, not a production-grade server.
Sum two integers as a decimal string.
Add two signed integers and return the sum. Result is a single text content block holding the sum formatted as a base-10 decimal string (e.g. `42`, `-7`).
Return the `text` argument unchanged.
echo
Return the `text` argument unchanged. Result is a single text content block whose body is byte-identical to the input (no quoting, no formatting, no trimming).
fixed sleep
Duplicate tool definitions with inconsistent descriptions. 'echo' and 'add' appear twice each with varying quality. The third 'echo' has only the name 'echo' as description (0 chars). This violates single-responsibility and confuses LLM tool selection.
No parameter descriptions visible. Input schemas show property names (text, a, b) but no descriptions explaining what each parameter controls. LLMs cannot infer parameter meaning from names alone.
Trivial or missing tool descriptions. 'echo' (5 chars), 'fixed sleep' (11 chars), and bare 'echo' (0 chars) fall far below the 194-char baseline. Descriptions under 20 chars cannot guide LLM tool selection.
No output schemas documented. LLMs cannot plan downstream tool calls or extract structured data without knowing what fields to expect in responses.
Inferred effective spec: 2025-06-18+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 38 | 2025-06-18+ | v2 |
No error handling guidance. Tools provide no recovery hints, error classification, or actionable messages. An LLM receiving a failure has no guidance on retry, user correction, or next steps.