writinglisted

# Writing Skill Iterative critique and improvement of long-form content. Parallel judges evaluate sections against acceptance criteria. User approves all changes. ## Architecture ``` USER: /cyber-writing <content-file> [--vision <file>] [--sources <files...>] │ ▼ PHASE 1: INGEST (main session) Read content + vision + sources Parse into sections (## headings) User picks scope (all / specific sections) Write /tmp/writing-brief.md │ ▼ PHASE 2: CRITIQUE (5 parallel subagents per section) ┌──────────┬──────────┬──────────┬──────────┬──────────┐ │PRACTICAL │ DATA │ LANGUAGE │SUBSTANCE │COMPLETE- │ │ JUDGE │FRESHNESS │ JUDGE │ JUDGE │NESS JUDGE│ │ │ JUDGE │ │ │ │ │"Would a │"Late-2025│"Sharp? │"Actual │"What's │ │CTO act │or 2026 │Concise? │insight │missing? │ │on this?" │data?" │No fluff?"│or common │What's │ │ │ │ │wisdom?" │excess?" │ └────┬─────┴────┬─────┴────┬─────┴────┬─────┴────┬─────┘ └──────────┴──────────┴──────────┴──────────┘ │ Each returns findings as task result │ ▼ PHASE 3: SYNTHESIS (main session) Deduplicate + consolidate judge outputs into: ├── PROBLEMS (what's wrong, with evidence + severity) ├── PROPOSALS (specific rewrites / additions / c