Skip to main content

The marketing-content skill

The reference skill the rest of the set builds on: a shared vocabulary for marketing copy, plus the channel-level best practices that decide what a piece of copy is doing. Pair it with anti-slop, which governs the prose itself; this one governs the structure and the strategy.

It's model-invoked: the agent consults it whenever it writes or reviews marketing copy, and the other skills (grill-me-content, repo-to-landing, content-architecture) use its terms so the whole set means the same thing by the same words.

Install​

The skills package is the default path:

npx @reactor-tools/skills add marketing-content

Or run npx @reactor-tools/skills with no arguments for the interactive picker (selecting any of the three workflow skills pulls this one in automatically - they reference its vocabulary).

Prefer the raw file? Download SKILL.md and place it yourself - e.g. for Claude Code:

mkdir -p .claude/skills/marketing-content
curl -o .claude/skills/marketing-content/SKILL.md https://docs.reactor.tools/marketing-content/SKILL.md

The vocabulary​

Six terms, used exactly:

  • Reader - the one person a surface is written for. Not a persona wall.
  • Hook - the earned first line: a specific claim, a surprising number, a named pain.
  • Promise - the one change in the reader's life this thing delivers. One per surface.
  • Proof - evidence shown rather than asserted. "Blazing fast" is a claim; "p95 went from 900ms to 210ms" is proof.
  • Objection - the specific reason this reader would bounce, answered where it occurs in the flow of the page.
  • CTA - the one action a surface asks for, verb-first. A page with three asks has none.
  • Message home - the single surface where a claim canonically lives; everything else links or shortens, never restates.

What else is in it​

The hierarchy of proof (show > cite > claim), the one-page-one-job rule, native structure sketches per channel (landing page, launch post, docs intro, email, X/LinkedIn), title and hook patterns that survive anti-slop, search-intent basics without keyword slop, and a measurement mindset (every CTA implies a metric).

It reads CONTENT.md and VOICE.md from your repo root when they exist - they outrank everything in the skill.

This is the manual version. Reactor runs the same discipline automatically against your saved brand voice, Guard rules, and performance data.