fusengine/agents

5 stars · Last commit 2026-04-05

Redefining development through cognitive automation and collaborative agent systems.

README preview

# Fusengine Claude Code Plugins

![version](https://img.shields.io/badge/version-v1.38.71-blue?style=flat-square) ![plugins](https://img.shields.io/badge/plugins-18-brightgreen?style=flat-square) ![agents](https://img.shields.io/badge/agents-19-blueviolet?style=flat-square) ![skills](https://img.shields.io/badge/skills-125-orange?style=flat-square) ![License: MIT](https://img.shields.io/badge/License-MIT-yellow?style=flat-square) ![platform](https://img.shields.io/badge/platform-macOS%20%7C%20Linux-lightgrey?style=flat-square) ![Windows](https://img.shields.io/badge/Windows-soon-orange?style=flat-square)

> A plugin ecosystem that turns Claude Code into a supervised, multi-agent development environment. Expert agents write code, hooks enforce quality in real-time, skills inject framework-specific knowledge, and **intelligent cartography auto-maps plugins and projects** — so Claude never guesses, never duplicates, and always follows your architecture.

![Statusline](docs/img/statusline.png)

## What It Does

**Without plugins:** Claude Code writes code based on its training data. It can hallucinate APIs, duplicate existing code, ignore your project structure, and produce inconsistent quality.

**With Fusengine plugins:**

- **Expert agents** detect your project type (Next.js, Laravel, React, Astro, Swift...) and load framework-specific documentation via MCP servers before writing a single line
- **82 hooks** intercept every Write/Edit/Bash call in real-time to enforce file size limits (<100 lines), block code duplication (DRY), require SOLID references, and validate security
- **125 skills** inject copy-paste-ready templates, architecture patterns, and best practices directly into agent context — no hallucination needed
- **APEX workflow** structures every task through Analyze → Plan → Execute (TDD) → Review → Validate — preventing the "just write code and hope" approach
- **Sniper validation** runs a 7-phase quality check after every modification: explore → research → grep usages → lint → fix → zero errors

View full repository on GitHub →