Skip to main content

The repo-to-landing skill

Turns a technical repository into landing-page copy where every claim carries a citation to the code that proves it. That one rule is the whole skill: capabilities the code can't evidence never reach the page, however good they'd sound - which is what makes the output shippable instead of aspirational.

Invoke it with /repo-to-landing (Claude Code, Cursor) or $repo-to-landing (Codex).

Install​

The skills package is the default path:

npx @reactor-tools/skills add repo-to-landing

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/repo-to-landing
curl -o .claude/skills/repo-to-landing/SKILL.md https://docs.reactor.tools/repo-to-landing/SKILL.md

How it works​

  1. Evidence sweep. The agent explores the repo (README, docs, manifests, CLI help, public API, examples, test names, changelog) and builds a proof inventory: what this thing verifiably does, one citation per capability. Anything the README asserts but the code can't show goes on a "claimed but unevidenced" list for you - never on the page.
  2. Positioning check. If CONTENT.md exists it's used; otherwise a short grilling round (at most 7 questions) settles reader, promise, why-now, objections, and the CTA. Genuinely unsettled positioning gets referred to a full grill-me-content session instead of guessed at.
  3. The draft. A standard landing skeleton from marketing-content - hero, 2-4 proof-backed pillars, how-it-works, objections answered inline, one CTA - emitted as Markdown with an explicit slot map (hero.title, pillar[1].proof, …) so it drops into any framework, each slot citing its evidence. Point it at existing page files and it writes into those instead.
  4. The audit. A citation walk (every claim back to its inventory entry, cuts for anything without one) plus the full anti-slop self-review.

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