mcp-cutup defines 4 tools with complete JSON schemas and basic descriptions. However, descriptions are terse (10-80 chars, well below the 194-char baseline), lack context on WHEN to use each tool, and omit error guidance. Tool names are domain-specific but lack action verbs (distill, fold, press, mark are nouns/verbs but not verb_noun patterns like 'reduce_text' or 'combine_texts'). Parameters have types and descriptions, but descriptions are minimal and lack format/constraint details. No output schemas documented. No error handling guidance. No tool annotations (readOnlyHint, etc.). The server is a creative text-manipulation tool, not a business-logic tool, so some naming latitude is acceptable, but descriptions remain below production baseline.

• Expand tool descriptions to 150 - 250 chars. Include: what the tool does, when to use it (vs similar tools), what it returns, and any prerequisites. E.g., 'distill: Reduces text length by removing words/letters deterministically. Use when you need a shorter version of text while preserving meaning. Returns the reduced text and reduction percentage. Seed ensures reproducible results across calls.'
• Document output schemas for each tool. E.g., distill returns {text: string, reduction_pct: number, original_length: number}. This enables LLMs to chain tools and extract results.
• Add error handling guidance to descriptions. E.g., 'If seed is invalid, returns error code INVALID_SEED with message 'Seed must be a non-negative integer'. Retry with a valid seed or omit for random behavior.'
• Rename tools to verb_noun patterns: distill → reduce_text, fold → combine_texts_folded, press → combine_texts_pressed, mark → encode_qr_code. This clarifies intent and aligns with production baselines.
• Add parameter constraints to descriptions. E.g., 'seed: Non-negative integer (0 - 2147483647) for reproducible reduction. Omit for random behavior.' and 'foldPct: Integer 1 - 100. Only applies when method is 'splice'; ignored for other methods.'