A shared collection of Claude Code plugins for the Sound in Theory development team.
Run the following slash command in Claude Code:
/plugin marketplace add soundintheory/claude-plugins
This registers the marketplace under the name soundintheory. You only need to do this once.
Browse available plugins:
/plugin
Or install directly by name:
/plugin install <plugin-name>@soundintheory
For example:
/plugin install example-plugin@soundintheory
/plugin update
- Create a new directory under
plugins/<your-plugin-name>/ - Follow the structure below
- Open a pull request — plugins are available to the team once merged to
main
plugins/
└── your-plugin-name/
├── .claude-plugin/
│ └── plugin.json # Required metadata
├── README.md # Plugin documentation
├── skills/
│ ├── your-skill/
│ │ └── SKILL.md # Model-invoked skill (context-aware guidance)
│ └── your-command/
│ └── SKILL.md # User-invoked slash command (/your-command)
└── .mcp.json # Optional MCP server configuration
{
"name": "your-plugin-name",
"description": "What this plugin does",
"author": {
"name": "Your Name",
"email": "you@soundintheory.co.uk"
}
}Model-invoked — Claude activates the skill automatically based on task context. The description field controls when it triggers:
---
name: your-skill
description: This skill should be used when the user mentions "X", asks about "Y", or is working on Z.
version: 1.0.0
---User-invoked — Creates a /your-command slash command the user can call explicitly:
---
name: your-command
description: Short description shown in /help
argument-hint: <required> [optional]
allowed-tools: [Read, Glob, Grep, Bash]
version: 1.0.0
---See example-plugin for a working reference implementation.