Writer

**Judgement:** This skill is a focused, rule-based post-processor for AI text that addresses many of the most common “LLM voice” issues (repetitive paragraph openers, monotone cadence, vague claims, list overuse, weak transitions, wordiness, and nominalizations). For teams fighting obviously machine-like style in their outputs, it is likely to provide tangible improvements with low overhead. **Strengths:** - Concrete, inspectable rules (e.g., checking whether 3+ paragraphs start with “This/The/It”, enforcing varied sentence lengths, replacing vague adjectives with numbers, and cutting stock wordy phrases). - Good coverage of stylistic traps that frequently appear in generic AI outputs, especially for marketing/essay-style text. - Encourages evidence, specificity, and concision, which typically align with better UX and higher perceived quality. **Risks / Limitations:** - Rule-based approach may overcorrect in some contexts, introducing unnatural variety or violating strict style guides (e.g., academic, legal, or highly formal writing that expects certain patterns). - Heuristics like opener checks or list reductions might conflict with user-intended structure, especially in technical documentation or step-by-step content. - No explicit adaptation to user voice; repeated application could gradually impose a particular stylistic fingerprint across different authors. **Recommended Scenarios:** - As a final polishing step for AI-generated marketing, content, and general prose where a more human, conversational feel is desired. - Within content pipelines that already accept moderate stylistic normalization in exchange for readability and engagement. - Less ideal for contexts where strict formatting, consistent terminology, or domain-specific style guides take precedence over “natural” rhythm (e.g., legal briefs, academic papers, regulatory docs).