afrexai-claude-code-production

What This Skill Does

The afrexai-claude-code-production skill provides a structured methodology and architectural framework for maximizing productivity with Claude Code. Instead of treating the AI as a general-purpose chat bot, this skill treats it as a production-grade engineer that operates based on a specific, project-level 'brain' file (CLAUDE.md). It focuses on reducing context-switching, optimizing token usage, and enforcing rigorous code quality through preemptive architecture definitions, standardized command sets, and TDD-first workflows.

Installation

To add this skill to your OpenClaw environment, execute the following command in your terminal:

clawhub install openclaw/skills/skills/1kalin/afrexai-claude-code-production

Ensure that your project directory is already initialized with a git repository, as the workflow relies on conventional commit standards and branch hygiene.

Use Cases

1. Greenfield Project Kickstarting: Rapidly scaffold a production-ready repository with pre-defined architectural patterns and strict type-safety rules that prevent technical debt from day one.
2. Context-Heavy Feature Development: Manage complex features in large codebases by maintaining a clean CLAUDE.md, allowing the AI to understand dependencies and coding styles instantly upon starting a session.
3. Refactoring & Technical Debt Reduction: Utilize the TDD (Test-Driven Development) workflows to systematically isolate legacy modules, write verifying tests, and refactor code without introducing regressions.
4. Monorepo Management: Leverage nested CLAUDE.md files to define package-specific behaviors, ensuring Claude handles distinct modules with appropriate context.

Example Prompts

1. "Initialize the CLAUDE.md file in this project root using the standard architecture template, configuring it for our current Next.js 15 and Drizzle stack."
2. "Before we start building the authentication service, run a health check against the current CLAUDE.md rules and report any discrepancies in our existing file structure."
3. "Refactor the user profile controller to use the Result type pattern as defined in our architecture rules. Write the tests first, then update the service."

Tips & Limitations

• Token Management: Keep your CLAUDE.md concise (under 200 lines). Bloated documentation increases latency and reduces model accuracy.
• Regular Audits: Run the 'Quick Health Check' table provided in the documentation before every session. If your score is below 6, stop and fix your context before writing code.
• Git Hygiene: This workflow assumes you are using git. Always commit or stash work before starting a major task to allow the AI to perform diff reviews safely.
• Human-in-the-loop: The system is designed to be an assistant. You must verify all diffs generated; do not blindly accept large refactors without reading the output.