wednesday-solutions/ai-agent-skills

Registry

Summary

A curated collection of pre-configured agent skills for Vibe Coded projects, designed to guide AI coding assistants like Claude Code and Cursor with specific code quality and design standards.

Key features

• Pre-configured skills for AI coding assistants
• Supports Claude Code, Cursor, and Copilot
• Focus on code quality and design standards
• Tailored for Vibe Coded projects
• Frontend and product engineering skills included

Use cases

• Enforce consistent code quality in Vibe Coded projects
• Guide AI assistants with design standards
• Streamline frontend development workflows
• Standardize AI-generated code across teams

README excerpt

# Wednesday Agent Skills Transform any repository into an **AI-aware, intelligent environment** — a codebase that any AI agent can jump into on day one without making junior-level mistakes. ```bash npx @wednesday-solutions-eng/ai-agent-skills install ``` --- ## What This Is Most AI agents in large codebases fail in the same ways: they hallucinate structure, waste tokens re-reading files they've seen before, and make changes without knowing what will break. This system solves all three. It works by pre-computing a **structural dependency graph** (SQLite, AST-based, zero LLM) of your entire codebase once, then giving every AI agent — Claude Code, Cursor, Gemini CLI, GitHub Copilot — a precise manual for how to work in *your specific project*. From that point on, structural questions are answered from the graph in milliseconds, not from re-reading source files. ## What You Get After Install This system integrates directly with **Claude Code**, **Gemini**, and other AI tools as an **on-demand skill**, feeding them structured repo context and enforcing local guardrails. **Answering structural questions via our pre-computed graph saves 70–90% of LLM tokens per query.** ### 1. Unified Codebase Intelligence Hub (`.wednesday/`) - **AST-parsed Dependency Graph (`graph.db`)**: Builds and maintains a SQLite database mapping every import, export, and function call across JS, TS, Python, Go, Swift, etc. This eliminates the need for AI agents to repeatedly read files, reducing LLM token consumption by up to 90%. - **Living Agent Configuration Files**: Automatically generates and updates `CLAUDE.md`, `GEMINI.md`, `.cursorrules`, and `.github/copilot-instructions.md` with available skills and rules. ### 2. Guardrails and Enforced Quality Standards - **Complexity Gates**: Enfor

Topics