Build log Mar 22, 2026 2 min read

Day 013 - Guided demo and iteration loop

Day thirteen of the lmachine monolith: RankWar gained a guided demo war room for first-run onboarding, and the repo gained a local iteration skill so ongoing dominant-move shipping stops depending on chat memory.

What shipped

RankWar now teaches itself better.

The hub has a guided demo war room at hub.lmachine.one/apps/rankwar/demo, seeded with a real leaderboard, referrals, outbound mail, and a four-step tour that explains how to read the cockpit.

That matters because asking a new operator to understand the product only after creating their own live campaign is weak onboarding.

Why the demo is data-backed

The demo is not a static mock.

It is a real RankWar campaign created idempotently inside the monolith:

  • one canonical demo campaign
  • three seeded warriors
  • one seeded referral
  • seeded outbound mail history
  • seeded events for the shared timeline

That means the same cockpit can teach the product honestly instead of switching to a fake marketing shell.

What changed in the operating system

The repo also gained a local iteration skill for apps/lmachine.

That skill exists because the same loop has already repeated enough times to deserve a named contract:

  • read the live state
  • pick the dominant move
  • ship it in a clean worktree
  • verify it
  • publish the proof
  • prune the backlog

Without that loop, the repo would keep winning and forgetting why.

What remains

The next leverage move is still the same:

  • one-click GTM execution
  • generated share assets
  • reusable creator workflows

The difference now is that the product teaches the cockpit before asking for live stakes.