Writing Tools MCP provides 5 well-intentioned text analysis tools with mostly complete descriptions and input schemas. However, there are significant gaps in parameter documentation, output schema clarity, and error handling guidance. Tool naming follows good verb-noun conventions (e.g., character_count, spellcheck, readability_score), but several parameter descriptions lack specificity about valid ranges, formats, and expected behavior. The readability_score tool's 'level' parameter includes an enum, which is good; however, other parameters lack similar constraints. Output schemas are not explicitly documented anywhere in the provided source, forcing LLMs to infer structure. Error handling is absent, no guidance on what to do if spellcheck fails, if text is empty, or if readability analysis encounters malformed markdown. The server lacks tool annotations (readOnlyHint, idempotentHint) despite all tools being read-only operations.

• Document the return type for each tool in structured output schema format. For character_count, specify: {"count": <integer>}. For spellcheck, specify: {"misspelled_words": [<string>], "suggestions": {<word>: [<string>]}}. For readability_score with level='full', specify: {"flesch_reading_ease": <number>, "flesch_kincaid_grade": <number>, "gunning_fog": <number>}, and document the nested structure when level='section' or 'paragraph'.
• Add error handling guidance to each tool description. Example for spellcheck: 'If the text is empty or under 20 characters, returns {"misspelled_words": []}. If spaCy language model fails to load, returns error with message: spacy_model_load_failed, ensure en_core_web_sm is installed.' Similar additions for all tools.
• Enhance the 'text' parameter description with concrete constraints. Example: 'The input string to analyze. Must be 1 - 100,000 characters, UTF-8 encoded. Supports mixed-case text; punctuation and special characters are preserved.'
• Add readOnlyHint tool annotations for all 5 tools since none of them modify server state. This helps agents understand tool safety and plan accordingly.
• For readability_score, explicitly list output fields in the description: 'Returns Flesch Reading Ease (0 - 100, higher = easier), Flesch-Kincaid Grade (US school grade level), and Gunning Fog Index (years of education). When level=section or level=paragraph, returns an array of scores, one per section/paragraph.'
• Remove or clarify the 'preprocess_text settings' reference in spellcheck. Either expose a preprocessing_level parameter (e.g., enum [minimal, standard, aggressive]) or state: 'Uses standard preprocessing: converts to lowercase, removes punctuation, filters stopwords.'

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry