AIAny
Icon for item

claude-code-best-practice

Collects Claude Code commands, agents, skills and engineering best practices with ready CLAUDE.md templates and orchestration examples. Focuses on reusable agent workflows, hooks, and MCP integrations for productionizing Claude-based coding/automation.

Introduction

Most repos that claim “best practices” are just curated tips; this one is organized as a living, example-driven toolkit that maps Claude Code concepts (agents, commands, skills, hooks, MCP servers) to concrete files and orchestration patterns you can drop into a project. It collects templates, workflow patterns, and implemented examples that show how to turn prompt patterns into reusable assets while preserving workspace safety and memory strategies. (github.com)

What Sets It Apart
  • Concrete, repo-first approach: instead of an essay, the project provides file-layout conventions (.claude, .mcp.json, CLAUDE.md), sample agents/commands/skills, and runnable orchestration examples so teams can copy patterns directly into their codebase — this reduces the guesswork between reading guidance and implementing it. (github.com)

  • Focus on agentic engineering and context management: includes patterns for persistent memory, context forking, sandboxing, and checkpointing (git-based rewind) so that workflows remain auditable and recoverable in long-running sessions. This addresses real integration concerns when Claude interacts with files, actions, or external MCP services. (github.com)

  • Community-driven and example-rich: the repo tracks community contributions, videos, and changelogs (active commits and issue/PR activity), making it useful both as a reference and as a living collection of evolving best practices. (github.com)

Who It's For — and Tradeoffs

Great fit if:

  • You use Claude/Claude Code (Anthropic) or similar agentic tooling and want production-ready templates (CLAUDE.md, MCP settings) you can copy into repositories.
  • You maintain a team workflow that benefits from encoded commands, subagents, and reusable skills rather than ad-hoc prompts.

Look elsewhere if:

  • You only need one-off prompt examples (this repo is structured around integrating patterns into repos and workflows).
  • You require vendor-neutral, model-agnostic SDKs — while many patterns are portable, several entries are explicitly scoped to Claude Code concepts and file conventions. (github.com)
Quick provenance note

The project owner and earliest commit author is Shayan Rais; the repository’s initial commit metadata shows the repository content created on 2025-10-31 (initial commit). The README and file tree provide the full list of included examples, topics, and implementation folders referenced above. (github.com)

More Items

GitHub

Hands-on lecture series that teaches neural networks from first principles up to building a GPT: each lecture pairs a YouTube video with Jupyter notebooks and exercises so you code models (micrograd → MLPs → WaveNet-like convs → GPT) while learning training and debugging.

GitHub

Hands-on, phase-based curriculum for building end-to-end AI systems from first principles — implement algorithms, run tests, and ship reusable artifacts (prompts, skills, agents, MCP servers) across Python, TypeScript, Rust, and Julia under an MIT license.

GitHub

A step-by-step, beginner-first programming course that teaches 'vibe coding'—conversational workflows to turn ideas into AI-enabled web and full‑stack prototypes. Features interactive simulated coding, multi-language docs, stage-based projects (from simple demos to SaaS capstones) and advanced agent/Claude Code guidance.