Vibe coding skills for Claude Code — TDD, debugging, planning, subagents, context management, and more.
npx claudepluginhub andrewchang-cpu/vibe-coding-configurationCore development skills for day-to-day project building with Claude Code.
Bootstrap a consistent AI agent environment in any project. Run once to generate config files for Claude Code, Cursor, and Codex — built around your project context.
npx vibe-setup
Prompts you to select which tools to configure. Or use flags:
npx vibe-setup --all # set up everything
npx vibe-setup --claude # Claude Code only
npx vibe-setup --cursor # Cursor only
npx vibe-setup --codex # Codex only
npx vibe-setup --claude --cursor # mix and match
--claude setup)Set the following as environment variables on your machine. The setup script reads them at install time; no .env file is required.
| Variable | Where to get it |
|---|---|
GITHUB_PERSONAL_ACCESS_TOKEN | github.com/settings/tokens |
CONTEXT7_API_KEY | context7.com |
TAVILY_API_KEY | tavily.com |
macOS/Linux — add to ~/.zshrc or ~/.bashrc:
export GITHUB_PERSONAL_ACCESS_TOKEN=your_token
export CONTEXT7_API_KEY=your_key
export TAVILY_API_KEY=your_key
Windows — set as User Environment Variables (search "Edit environment variables" in Start), or via PowerShell:
[System.Environment]::SetEnvironmentVariable("GITHUB_PERSONAL_ACCESS_TOKEN", "your_token", "User")
[System.Environment]::SetEnvironmentVariable("CONTEXT7_API_KEY", "your_key", "User")
[System.Environment]::SetEnvironmentVariable("TAVILY_API_KEY", "your_key", "User")
A .env file in your project root also works as a fallback, but environment variables take precedence.
| Tool | Files |
|---|---|
| Claude | CLAUDE.md, .vibe/ |
| Cursor | .cursor/mcp.json, .cursor/rules/, .vibe/ |
| Codex | AGENTS.md, .codex/config.json, .vibe/ |
Re-running is safe — .vibe/project-context.md is never overwritten. All other files are regenerated.
npx vibe-setup.vibe/project-context.mdnpx vibe-setup to regenerate all config files with your project context baked in.vibe/project-context.mdnpx vibe-setupI need you to analyze this codebase and populate `.vibe/project-context.md`.
This file drives all AI agent config — be specific and accurate, as it will be
read on every future session.
Work through the following steps in order:
1. **Explore the codebase**
- Read the directory structure (top-level and key subdirectories)
- Read package.json, any lockfiles, config files (tsconfig, eslint, etc.)
- Identify the stack: languages, frameworks, major libraries
- Understand the architecture: how is the code organized? What are the main layers/modules?
- Identify any existing conventions: naming patterns, file structure, import style
2. **Write `.vibe/project-context.md`** with the following sections:
**Project Context**
- Name: the project name
- Stack: the actual languages, frameworks, and key libraries in use
- Architecture: a concise description of the structure — what the major parts are and how they relate
**Project Architecture & Directory Map**
List the real folder structure with a one-line description of each key directory's purpose.
**Anti-Patterns & "Never Do This"**
Identify 3–7 patterns that would be easy to introduce but would break conventions, cause bugs,
or violate the architecture (e.g., bypassing an abstraction layer, mixing concerns, hardcoding values).
**Git & Workflow Standards**
Check for a CONTRIBUTING.md, existing commit history, or branch naming patterns. Document what
you find. If nothing exists, propose a sensible default.
**Definition of Done**
A checklist an agent must complete before marking any task done — make it specific to this project
(e.g., "passes existing test suite", "no new TypeScript errors", "changelog updated if applicable").
**Useful Project Commands**
The actual commands to run the dev server, build, and test suite.
Start with step 1 before writing anything. Show me the proposed content before writing to disk
so I can approve or adjust.
I need you to help me define a new project and write `.vibe/project-context.md`.
This file drives all AI agent config — it will be read on every future session.
Here's what I want to build:
[DESCRIBE YOUR PROJECT HERE — what it does, who it's for, any tech preferences or constraints]
Work through the following steps in order:
Official prompts.chat marketplace - AI prompts, skills, and tools for Claude Code
Open Design — local-first design app exposed to coding agents over MCP. Install once with your agent's plugin command and projects/files/skills are reachable through stdio.
Behavioral guidelines to reduce common LLM coding mistakes, derived from Andrej Karpathy's observations