wshobson/agents

This repository is a large collection of plugins, agents, and skills designed to extend Claude Code, Anthropic's AI-powered coding assistant. The problem it addresses is that Claude Code on its own is a general-purpose tool, but real software development involves many specialized domains: security scanning, Kubernetes deployments, machine learning pipelines, API design, test automation, and more. Keeping all that knowledge loaded at once would be wasteful and slow. This project provides a structured marketplace of 80 focused plugins, each covering one specific domain, so you load only what you actually need for a given project. The architecture works through a plugin system built into Claude Code. You first register the marketplace with a single command, then browse available plugins and install the ones relevant to your work. Each plugin packages a set of AI agents (specialists with domain-specific instructions), agent skills (modular knowledge packages that load only when invoked), and slash commands (shortcuts for common tasks). For example, installing the Python plugin brings in a Python expert agent, a Django expert, and a FastAPI expert, plus sixteen skills for things like writing tests, generating type hints, or optimizing database queries. A multi-agent workflow plugin can coordinate several of these specialists in sequence, such as running an architect review, a code review, and a security audit in parallel on a pull request. You would use this project if you want Claude Code to act as a deep specialist rather than a generalist: have it take ownership of scaffolding a new microservice, harden a Kubernetes cluster, or lead a full incident response workflow while keeping Claude's context window efficient. It is a Python-based project targeting Claude Code users who want production-grade automation across complex software engineering tasks.