My Claude Code Setup 🤖 Agent Open source

What it does

My Claude Code Setup is a starter configuration for Claude Code projects, shared by the author of the Centmin Mod project. Its centrepiece is a "memory bank" system built on CLAUDE.md: a set of memory files and instructions designed to keep context across many chat sessions, so the agent retains project knowledge instead of relearning it each time. Around that it collects settings, hooks, slash commands, subagents, skills and status lines you can copy into your own project and trim to taste. It is meant as a template to fork and adapt, not a packaged product.

What is inside

• Three modern CLAUDE.md templates following Anthropic's guidance (progressive disclosure, dual-memory architecture) plus a migration prompt for existing files
• A statusline script that shows model, context-window usage, cost metrics and git branch
• Git worktree launcher scripts for isolated coding sessions on macOS, Linux and Windows, and instructions for running Claude Code against Z.AI's GLM models
• Custom subagents such as a memory-bank synchroniser, a code searcher and a datetime helper
• Slash-command groups for cleanup, documentation, security, architecture, refactoring and prompt engineering
• Notes on settings, permissions, MCP servers (Context7, Notion, Chrome DevTools and others) and third-party plugin marketplaces

Works with

Claude Code, primarily. The .claude/settings.json uses macOS Terminal-Notifier and can be removed on other systems. Some sections also cover Codex, Gemini CLI and Copilot CLI.

How to install

There is no one-line installer. Copy the repository files into your project directory, edit CLAUDE.md and the templates to fit, then run /init so Claude analyses your codebase and populates the memory-bank files. The README suggests faster search tools:

brew install ripgrep fd jq

Maintenance and safety

MIT-licensed and updated recently. It contains affiliate and discount links (for example to Z.AI), which are marketing rather than part of the tooling. Read each hook and command before enabling it, and drop the parts you do not need so they do not consume context.

Who should use it

Developers setting up a new Claude Code project who want a maintained, opinionated starting point for CLAUDE.md, hooks and commands, especially those interested in a persistent memory-bank workflow.