Multi-language MCP server implementation with tools for mathematical operations, Terraform validation, CSV data processing, and badge generation
Scoring was not performed
Two tools share identical name 'square_root' with different parameter schemas ('n' vs 'number'). This violates composition rules and forces ambiguous tool selection by LLMs.
ZERO parameter descriptions across all 5 tools. LLMs cannot infer what 'n', 'text', 'code', 'number' expect without explicit guidance. Violates critical requirement: 'Every parameter needs a description explaining what it controls.'
generate_svg tool lacks description of output format. Agents cannot plan next steps, is result a base64 data URI, SVG string, or file path? No guidance on when to invoke this tool vs alternatives.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 22 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 29 | - | v1 |
No output schemas documented for any tool. LLMs cannot predict return types, structure, or fields needed for downstream composition. Violates 'Document the output schema' requirement.
Tool descriptions below baseline. generate_svg (32 chars), iris_csv (27 chars), square_root (26 chars) fall well below 194 char average and lack context on WHEN to use, WHAT happens, or prerequisite steps.
No error handling guidance. tf_validate returns 'true or error list' but LLMs have no instruction on error format, retry strategy, or recovery path. Violates pattern:recovery-guide.
Naming convention inconsistency. iris_csv (noun-first), generate_svg (verb_noun), tf_validate (verb_noun), square_root (verb_noun). Not all follow action-verb-first pattern. iris_csv should be 'get_iris_data' or 'list_iris_records'.
Ambiguous parameter names without context. 'n' (iris_csv, square_root) does not indicate whether it is a limit, count, or offset. 'code' (tf_validate) lacks format specification (raw HCL vs JSON vs string with newlines).