Community
Community Spotlight
Real projects built by real builders. See what the community has shipped using AI-accelerated workflows.
2 projects
Vibecuterie
Vibecuterie is the curated home for vibe coded projects. Discover what people are building with ai, vote on your favorites, and find your next spark of inspiration. Submit your project to let others decide if you actually have taste.
Riley Beresini
Marketer
Expert-Level Claude Skill-Builder
Claude's default Skill builder is good... but this meta-skill will build you production-grade Skills with advanced architecture patterns baked in. Architecture patterns it implements: PRD-STYLE SPECIFICATION: Create Skills that act as mini product specs with (1) purpose - what it is and isn't for, (2) users - who uses it, their constraints, (3) non-goals - explicit boundaries. ADAPTIVE RECOMMENDATIONS: Don't force complex patterns on simple tasks. Create lightweight structures for yes/no tasks (run tests, check files) and robust validation only when quality varies. SCHEMA THINKING: Have Skills output two formats: (1) structured data (JSON) that other Skills can read (2) plain language explanations for you. This allows Skills to chain together smoother. SKILL-SPLITTING DETECTION: Catch when you're trying to build one Skill that should be multiple Skills chained together. This prevents messy multi-purpose Skills and makes it easier to troubleshoot. FAILURE MODE HANDLING: Build in backup plans so Claude doesn't go off the rails or stop if it's uncertain. SELF-UPDATING ARCHITECTURE: Create a process for the Skill to remember when you correct it. The next time you run it, those fixes apply automatically. It'll learn from your feedback without you needing to constantly rebuild. MEMORY-AWARE PATTERNS: Have it save additional files like your preferences (tone, format, context) so you're not constantly reexplaining. MULTI-PASS SYSTEMS: Build in processes to diagnose the input/output (what's working or not working), then adjust output based on criteria. This cuts down manual steering and feedback. EVALUATION CRITERIA: Turns varying, subjective judgment into specific rankings you can track and improve. VALIDATION SCRIPTS: Add Python scripts for automated output, quality checks, and metrics. DECOMPOSITION-FIRST: For complex multi-step tasks, break down the Skill workflow into subtasks, align on the plan, get approval, then execute. This prevents missed requirements.
Riley Beresini
Marketer
Built something cool?
Share your project with the community. We feature the best submissions in our newsletter and right here on this page.
Submit Your Project