MCP server providing a library of reusable prompts across multiple categories including code, writing, and analysis
This is a prompt library server, not a tool/function calling server. The 18 'tools' are actually MCP prompts that return templated text to guide LLM behavior. This mismatch is foundational: MCP prompts are stateless text generators, not executable actions. For a PROMPT server, the definitions are reasonable, each prompt has a name, description, and documented arguments. However, scoring against the tool-focused rubric reveals significant gaps. Argument schemas lack explicit type declarations (only name/description/required boolean visible). Descriptions are present but brief (40-60 chars typical). No output schema is documented. No error handling guidance. The server correctly implements list_prompts() and get_prompt() handlers and exports arguments as types.PromptArgument, but the underlying prompt definitions (ANALYSIS_PROMPTS dict) store arguments as plain dicts without JSON Schema type fields. All 18 tools/prompts share the same quality issues: minimal argument descriptions and no type constraints. The code is well-structured and readable, but tool quality is below A+ baseline for either tool or prompt domains.
Generate unit tests for code
Design or review API architecture and endpoints
Generate engaging blog posts on any topic
Analyze bug reports and prioritize issues
Comprehensive code review with best practices and security analysis
Analyze competitors and market position
Analyze data and provide insights
No JSON Schema type declarations on arguments. Arguments stored as plain dicts with 'name', 'description', 'required' keys only, missing 'type' field. This violates schema completeness and prevents structured validation/discovery.
No output/return schema documented. Prompts return only role='user' + TextContent(text=...). No schema declares what fields the generated prompt text will contain or how it is structured for downstream use.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 47 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 0 | - | v1 |
Design or review database schema
Debug code and find solutions to errors
Create technical documentation
Craft professional or casual emails
Explain complex code in simple terms
Generate meaningful git commit messages
Analyze and optimize code performance
Refactor code for better structure and maintainability
Perform security audit on code
Generate social media content
Perform SWOT analysis
Argument descriptions are brief and generic. E.g., 'The code to review' (18 chars), 'Programming language (e.g., python, javascript, java)' (53 chars). Baseline is 72 chars average. Many descriptions lack context about expected format, valid values, or dependencies.
No enum constraints on arguments that should be restricted. E.g., 'language' in code-review and 'framework' in add-tests accept free-form strings. Should be enums (python|javascript|java|etc., pytest|jest|junit|etc.) to guide LLM selection.
No error handling or recovery guidance. If an argument is missing or invalid, server will raise ValueError('Unknown prompt: {name}'). No guidance on valid prompt names, retryable errors, or what to do next.