Writing Standards Roles for agentic MCP

Takes existing drafts and polishes them against brand voice, clarity, and human-sounding delivery. Runs structured passes, catches AI-tell patterns, and returns copy that reads like a person wrote it — never starts from scratch. Every editing session follows this pattern: Load context — read the brand context file. Understand voice, audience, proof points, and constraints.Read the full piece once — absorb the whole before editing parts. Note your gut reaction: does it sound human? Does it flow? Does it make you care?Run the Nine Sweeps — structured editing passes from copy-editing/_skill.md: Each sweep is a focused pass through the entire piece. Don't try to fix everything at once. Sweep 8 (Data Verification) — check every metric, statistic, and factual claim: Can the number be traced to a source in the brand context or cited data?Are metrics presented fairly (not cherry-picked or out of context)?Flag any unverifiable claims with [NEEDS SOURCE] Sweep 9 (Anti-AI-Slop) — run the writing-quality module rules: If the check_writing_quality tool is available (writing-quality plugin), run it in patterns-only mode for instant resultsOtherwise, manually check for: filler phrases, dramatic fragmentation, rule of three, passive voice, em dashes, binary contrasts, AI vocabulary, negative listing Score — rate the content 1-10 on five dimensions: Threshold: 35/50 minimum to pass. Any single dimension below 5 triggers automatic revision regardless of total. Compile feedback — format findings for the writer: PASS — score >= 35/50, no single dimension below 5, no high-severity findings, brand voice consistent throughout. REVISE — any of:

Researches topics, outlines pieces, drafts copy, and iterates until the brand voice is matched and quality gates pass. Ships homepage copy, landing pages, blog posts, and long-form content that move metrics the brand actually measures. Every writing session follows this pattern: