What problem does it solve? Setting up the stuc-stack Android/Kotlin agent environment requires manually checking that upstream skill plugins (chrisbanes/skills, android/skills, the android CLI) are installed and deciding which AI model each workflow role should use. This Skill automates both: it doctors upstream dependencies and writes a persistent per-role model configuration rule. ## Core Features & Use Cases - Upstream doctoring: Probes for the android CLI binary, the android-cli skill, chrisbanes skill clusters, and android/skills leaves, printing exact install commands when anything is missing and refusing to claim the stack is ready. - Per-role model configuration: Detects available model slugs, applies a reasoning budget (unlimited/large/medium/small), and writes ~/.cursor/rules/stuc-stack-models.mdc mapping roles like arena runners, swarm workers, and interrogate reviewers to models. - Harness association: Links project-local docs/verify-* skills to the current agent harness (Cursor, Claude Code, or Codex) via thin delegate files. - Use Case: After cloning stuc-stack into a new project, run the setup to confirm chrisbanes and android skills resolve, pick a medium reasoning budget, and have every role's model rule written idempotently for future sessions. ## Quick Start Run /setup-stuc to check my upstream installs and configure per-role models with a medium reasoning budget.
