Project scaffolding and template upgrades require consistent choices and careful review of existing code. This skill guides vibestart CLI discovery, previews, capability additions, and maintenance.

Types and documentation may not explain unexpected dependency behavior. This skill guides AI agents to fetch matching source versions and inspect implementations through a reusable cache.

Unity UI Toolkit development requires careful handling of UXML structure, USS restrictions, and runtime interactions. This skill guides focused edits and generates layouts, custom visuals, and requested C# behavior using documented project patterns.

Unity projects can mix UI frameworks, making implementation choices unclear. This skill explains framework differences and routes requests using existing project patterns.

Recreating a screenshot can introduce unwanted design changes and visible mismatches. This skill guides your agent to implement the reference and compare rendered results.

Website business details can differ from a Google listing, making structured data harder to maintain. This skill builds JSON-LD from available listing fields and proposes an exact website diff for approval.

Web interfaces can block users who rely on keyboards, screen readers, or zoom. This skill guides targeted fixes and reports WCAG 2.2 criteria with verification limits.

Building browser-based 3D scenes requires coordinating rendering, assets, interaction, and performance. This skill provides Three.js examples, reference guides, and a WebGL starter to support implementation.

Cloudflare K2 setup and troubleshooting require current API details, configuration rules, and delivery semantics. This skill directs your assistant to relevant documentation before providing guidance.

Authentication setup and session handling can make end-to-end tests unreliable. This skill guides Clerk testing through framework-specific documentation, development keys, testing tokens, and session reuse.

Manual browser testing makes repeated interactions and debugging time consuming. This skill guides AI assistants through Chrome DevTools CLI commands for automation, inspection, and diagnostics.

Astrale Domain development spans Schema, authorization, runtime, and deployment, which can become inconsistent. This skill guides focused changes and phased creation with explicit verification boundaries.

Vague commit messages make change history difficult to understand. This skill uses staged changes to draft typed subjects and optional bodies explaining the reason.

Pull request reviews can miss issues across security, tests, performance, and maintenance. This skill organizes twelve review domains and prepares findings for GitHub comments.

Reviews can miss failure conditions or accept fixes without reliable evidence. Gauntlet guides scoped adversarial reviews, authorized repairs, and revision-bound verification within a finite budget.

Les projets TypeScript dérivent souvent vers des types faibles, des schémas dupliqués et des modèles d’état fragiles. Cette compétence guide Claude, Codex et Claude Code vers des modèles TypeScript plus sûrs.

Les projets Tamagui peuvent souffrir de tokens incohérents, de composants lents et de patterns inaccessibles. Cette skill fournit des conseils pratiques v1 pour la configuration, l’optimisation du compilateur, les formulaires, les overlays, les animations et les thèmes.

Les projets Python dérivent souvent vers un typage faible, des modèles mutables et une gestion des erreurs peu claire. Cette compétence donne à Claude, Codex et Claude Code des règles ciblées pour des modifications Python plus sûres et plus claires.

Les tests Playwright deviennent souvent instables, lents et bruyants dans les workflows d’agents. Cette skill fournit des modèles concis pour les locators, les assertions, les fixtures, la réutilisation de l’authentification, les mocks réseau et les reporters compacts.

Les projets Go peuvent dériver vers des types faibles, des interfaces peu claires et une gestion des erreurs incohérente. Cette compétence donne à Claude, Codex et Claude Code des conseils ciblés pour une conception Go idiomatique, les erreurs et la journalisation.