writing

name

writing

description

Use this skill any time the user wants to write, draft, edit, proofread, restructure, or improve prose — emails, proposals, reports, RFCs, vision documents, blog posts, slack/notion posts, or any text where structure and concision matter. Also triggers when the user asks for "Pyramid Principle" structuring, BLUF rewriting, marginal-utility editing, "make this shorter", "tighten this", "fix the structure", or supplies an existing draft and asks for review. Loads writer + editor personas and optional DVQ/RFC templates; runs a 3-phase Plan → Write → Edit pipeline with up to 3 revision cycles.

Writing Skill

Cowork-adapted version of the Pyramid Principle + Marginal Utility writing pipeline. The pipeline assumes the user wants production-grade business writing, not casual chat.

Argument Parsing

Treat the user's message as the task. If it starts with the keyword dvq or rfc, load the matching template before Step 1:

• dvq [topic] — Anthropic-style strategic vision document
• rfc [topic] — technical decision document

Strip the keyword from the topic before passing it on.

Preparation

Read these files into context before drafting:

1. personas/document-writer.md — writer persona
2. personas/marginal-utility-editor.md — editor persona
3. (if template keyword detected) templates/dvq.md or templates/rfc.md

Workflow

Execute three steps sequentially. Delegate each to an Agent (general-purpose subagent) so personas are isolated and the steps don't bleed.

Step 1: Plan (structure design)

Subagent prompt includes the writer persona. Instructions:

• Identify reader: who they are, what they already know, what decision/action this document supports
• If a template was loaded, use its structure as the starting point
• Design structure per Pyramid Principle: conclusion (1 sentence) → arguments (MECE-grouped) → evidence
• Verify each argument answers "why?" or "how?" from the conclusion
• Hierarchy ≤ 3 levels
• Decide format (short / medium / long)

Step 2: Write (drafting)

Subagent prompt includes writer persona + Step 1 output. Instructions:

• Write per the structure
• Conclusion first
• Topic sentence at the head of each paragraph
• Narrative prose; bullets only when they aid comprehension
• Concrete numbers and facts instead of adjectives/adverbs
• Output the draft only — no meta-commentary

Step 3: Edit (marginal utility)

Subagent prompt includes editor persona + Step 2 draft. Instructions:

• Verify Pyramid structure (conclusion-first, topic sentences, MECE, ≤ 3 levels)
• Apply marginal utility test against the identified reader
• Expression: adjective → number, passive → active, drop "you can"/"there is" padding
• Reader-level adaptation: if reader is non-technical, flag undefined technical terms on first use
• Scannability: paragraphs ≤ 5 sentences; headings carry the conclusion word; lists keep parallel grammatical structure
• Volume: body ≤ 6 pages; excess to appendix
• Output: editing report + revised draft

Cycle Decision

If editor returns "revision needed", loop back to Step 1 with the editor's feedback. Cap at 3 cycles. If the cap is hit, output the best draft so far as final.

Final Output

Present the editor-approved (or final-cycle) draft to the user.

When NOT to use

• Single-sentence rewording
• Code comments (use the project's coding conventions instead)
• Casual chat replies
• Translation-only tasks (Cowork has no separate translation skill, but the 3-phase pipeline is overkill)