Hivemind AI

A Claude Code marketplace for agentic plugins, skills, and workflows.

Overview

Hivemind AI provides modular, composable plugins that enhance Claude Code with specialized capabilities:

• Subagents: Task-specific AI agents for focused work
• Skills: Domain knowledge modules (lazy-loadable documentation)
• Hooks: Runtime enforcement and workflow automation
• Commands: User-facing workflow orchestration

Requirements

• Bun - Required for plugin hooks (phase enforcement, queue processing)

  curl -fsSL https://bun.sh/install | bash

Installation

# Add marketplace
claude plugin marketplace add git@github.com:hivemind-ai-core/claude-hivemind-ai.git

# Install plugin
claude plugin install superagents@hivemind-ai

Available Plugins

superagents

Full-featured Research-Plan-Implement (RPI) workflow with Test-Driven Development (TDD).

Features:

• Hierarchical agent architecture
• Red-Green-Refactor TDD enforcement
• Context compression and management
• 14+ skills across multiple tech stacks
• Automated commit workflows

Quick Start:

# After installing superagents
/superagents:setup
/superagents:update-roadmap
/superagents:work

Tip: Type /work and press Tab to autocomplete to /superagents:work - works for all commands!

View superagents documentation

Plugin Structure

Each plugin follows a standard structure:

plugin-name/
├── README.md           # Plugin documentation
├── agents/             # Subagent definitions
├── skills/             # Domain knowledge modules
├── hooks/              # Runtime hooks
├── commands/           # User commands
└── .template/          # Setup templates

Creating Plugins

See the Plugin Development Guide for creating new plugins.

Philosophy

1. Composable: Plugins work independently or together
2. Focused: Each component has single responsibility
3. Lazy-loaded: Skills and context loaded on-demand
4. Transparent: File-based communication, human-readable
5. Evolvable: Learn from mistakes, accumulate patterns

License

MIT