The grill-me-content skill
A relentless interview that develops great content and copy for your site, then writes it from your answers. The premise: bad copy usually isn't a writing problem, it's an alignment problem - nobody pinned down the reader, the promise, or the proof. So the skill grills you until the copy can only come out right, and only then writes it.
Invoke it with /grill-me-content (Claude Code, Cursor) or $grill-me-content (Codex).
Install
The skills package is the default path:
npx @reactor-tools/skills add grill-me-content
Or run npx @reactor-tools/skills with no arguments for the interactive picker.
Prefer the raw file? Download SKILL.md and place it yourself - e.g. for Claude Code:
mkdir -p .claude/skills/grill-me-content
curl -o .claude/skills/grill-me-content/SKILL.md https://docs.reactor.tools/grill-me-content/SKILL.md
How a session runs
- Explore first. The agent reads what exists (README, docs, current marketing pages,
CONTENT.md/VOICE.md) so facts get looked up, not asked. Only decisions reach you. - The grilling. One question at a time, each led by the agent's recommended answer so you can accept it in a word. It walks the tree in order: reader → promise → proof inventory → alternatives and why-now → objections → voice → which surfaces to produce and their CTAs. Vague answers get grilled down: "developers" is not a reader, "it's fast" is not proof.
- The record. Decisions persist to
CONTENT.md(positioning, proof inventory, objections) andVOICE.md(register, lexicon, like-this/not-like-this pairs) at your repo root - so no future session re-asks, and every other skill reads the same answers. - The copy. Each agreed surface gets written using marketing-content structure under anti-slop enforcement, with every claim traceable to the proof inventory. No confirmed claim, no copy.
This is the manual version. Reactor runs the same discipline automatically against your saved brand voice, Guard rules, and performance data.