Back to directory

Claude Code Infrastructure Showcase

A reference library, not a working application, of production-tested patterns for Claude Code: auto-activating skills, hooks, specialized agents and a dev-docs system, drawn from six months of use on a TypeScript microservices project.

Claude Code Infrastructure Showcase

Claude Code Infrastructure Showcase is a curated reference library of production-tested Claude Code infrastructure, built from six months of real-world use managing a TypeScript microservices project. It is explicitly not a working application: a developer copies the patterns they need into their own project.

It documents three core building blocks: a skill, a markdown guide of patterns and conventions that Claude loads when relevant; a hook, a script Claude Code runs automatically at key moments such as every prompt or before and after edits; and skill-rules.json, the configuration that tells hooks which prompts or files should trigger which skills. Around these, it provides a pattern for auto-activating skills through hooks, a modular skill pattern that keeps each skill under 500 lines with progressive disclosure, specialized agents for complex tasks, and a dev-docs system meant to survive context resets.

Examples use a generic blog domain. It is written in TypeScript under the MIT license.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.