What problem does it solve? AI coding agents often produce verbose prose, unverified claims, and inconsistent architecture decisions on Android/Kotlin projects. This Skill enforces a disciplined working style: concise replies, named engineering principles, structured playbooks for common tasks, and mandatory runtime verification before declaring work done. ## Core Features & Use Cases - Named principles and playbooks: Routes tasks like bug fixes, features, refactors, perf work, and PR babysitting through step-by-step playbooks grounded in cited principles (e.g., Prove It Works, Model the Domain). - Verification loop: Requires Gradle assemble plus the android CLI to prove Android UI and device behavior on the real artifact, never accepting "it compiles" as done. - Subagent orchestration: Prescribes deliberate delegation with per-role models, parallel design exploration, and adversarial review for contested designs. - Use Case: Ask the agent to fix an Android crash in /stuc-mode; it reproduces the bug on a device via the android CLI, root-causes it, delegates the fix, verifies the repro now passes, and opens a PR with evidence. ## Quick Start Ask the agent to work in stuc-mode on your Android or Kotlin task, for example by saying "use /stuc-mode to fix this Compose recomposition bug and verify it on device".
