All updates

Agents

v2.5

September 1, 2026

The kit is agent-native now

AGENTS.md, first-party skills, /llms.txt, tool approvals, and visible reasoning so Cursor and other agents can operate the repo without guessing.

Buyers were already dropping the repo into Cursor on day one. V2.2 treats that as a first-class surface: the agent gets a map of the monorepo, skills for the jobs it actually does, a public /llms.txt, and a safer tool-approval path with reasoning you can see.

AGENTS.md + skills

A repo-level briefing and reusable skills so agents stop rediscovering the same commands every session.

/llms.txt

A public, machine-readable index of the product so agents can load context without scraping the marketing site.

Visible reasoning

Tool approvals and model reasoning stay on screen, so you can see what the agent is about to do before it does it.

Ship on V2 today

Get the full monorepo — web, mobile, extension, SDK, and MCP.

Zero To Shipped V2 gives you the production Next.js app, Expo mobile app, browser extension, auth, billing, REST API, and deploy paths already connected.

Start Shipping

One-time payment. Lifetime updates.

Built for the way people actually start

Most buyers open the repo in an agent first and only later read the docs. That only works if the repo explains itself.

  • AGENTS.md covers the monorepo layout, the commands that matter, and the traps that waste a first afternoon
  • Skills encode recurring jobs — env, migrate, ship, mobile — instead of hoping the model remembers
  • `/llms.txt` gives crawlers and agents a clean entry point
  • Tool approvals keep destructive actions behind an explicit yes

Skills, not folklore

The recurring jobs — env, migrate, ship, mobile — are encoded as skills so the next session does not rediscover them.

  • First-party skills for the jobs buyers actually run
  • Tool approvals stay on screen before anything destructive
  • Reasoning stays visible so you can interrupt a bad plan