Skills
Wire AI ships as Claude skills for coding agents. They cover both the integration and the ongoing improve loop, so most of the work happens inside your editor.
wire-rn-integrationFastest pathThe fastest way in. Point it at your app and it detects your conventions, wires Metro, derives a theme, scaffolds the onboarding screen, wires a fallback flow, and type-checks the result.
- When it fires
- When you first integrate the kit into a React Native or Expo app.
- How to use
- Run the wire-rn-integration skill from the root of your app.
wire-aiImprove loopDesigns the question script: the base questions with stable keys, the phased arc, and the screen budget for the 8 to 13 screen window. This is where you decide what the onboarding is actually trying to learn about a user.
- When it fires
- When you design or revise what the onboarding asks.
- How to use
- Invoke the wire-ai skill and follow section (c2), design the question script.
learning-reportImprove loopTurns a week of your funnel into an aggregate analysis, per-question abandonment, redundancy, personas, before and after, plus a paste-ready context update for the flow. The report is stored on the platform, so you see it as a card in the console.
- When it fires
- After about a week of real traffic.
- How to use
- Run the learning-report skill; it uploads via wireai_submit_learning_report.
No skill runner? The prompt fallback
If your agent cannot run skills, the kit ships INTEGRATION_PROMPT.md, a copy-paste prompt that does the same job by hand. You paste it into any coding agent from the root of your app, fill in three values at the top, your server URL, your wai_ app key, and your app id, and it walks the same ordered steps: detect your conventions, install the packages, wire Metro, add the env vars, build a theme from your brand, scaffold the screen with a fallbackFlow, mount it in your signup flow, and leave the result type-checking green.