This guide explains how to install and use plugins from this marketplace in Claude Code.
Install the marketplace once, then enable your plugin.
/plugin marketplace add RhombusSystems/claude-code-plugins
| Persona | Command |
|---|---|
| Developers building on Rhombus | /plugin enable rhombus-developer |
| Day-to-day Rhombus users | /plugin enable rhombus-user |
| MSP/reseller partners | /plugin enable rhombus-partner |
You can enable multiple plugins if needed.
Each plugin ships three kinds of capabilities. All three are live the moment you enable the plugin — no further activation needed.
Skills — model-invoked. Claude detects relevant user requests and applies the skill automatically. You don't type anything; just ask your question naturally ("how do I build a face recognition workflow on Rhombus?" fires the rhombus-api skill).
Slash commands — user-invoked. Type them directly in the prompt. Each plugin's README lists its commands, e.g.:
/rhombus-find-endpoint camera # developer
/rhombus-alerts "front lobby" 1h # user
/rhombus-fleet-report weekly # partner
Agents — model-invoked specialized subagents. Claude delegates to them when the request matches the agent's description (e.g., complex multi-step investigations, webhook debugging, client-selection).
Check each plugin's README.md for the complete list of skills, commands, and agents.
The rhombus-developer plugin auto-wires two MCP servers via its .mcp.json:
rhombus(official API MCP viarhombus-node-mcp) — requiresRHOMBUS_API_KEYexported in your shell.rhombus-docs(HTTP MCP atapi-docs.rhombus.community/mcp) — live doc search.
Verify with /rhombus-mcp-status.
If you prefer to embed skills directly in your project rather than using the plugin:
git clone https://github.com/RhombusSystems/claude-code-pluginsCopy any skill folder into your project's skills/ directory:
cp -r claude-code-plugins/plugins/developer/skills/code-review ./skills/Add the skill path to your project's Claude Code configuration so Claude knows it's available.
Trigger it in Claude Code the same way as a plugin-installed skill — ask Claude a matching question and it auto-applies.
- Skills (
SKILL.md) — Markdown files containing instructions Claude follows when the skill is active. Triggered by matching thedescriptionfield against the user's request. - Commands (
commands/*.md) — Slash-command entry points with frontmatter and a prompt body. Typed explicitly as/<command-name>. - Agents (
agents/*.md) — Subagent definitions with their own system prompt, tool set, and invocation triggers. - Hooks (
hooks/*.mdor.sh) — Event-driven automation (PreToolUse, SessionStart, etc.) that runs around tool calls and session events. - MCP servers (
.mcp.jsonat plugin root) — External tool integrations auto-attached when the plugin is enabled. Tools appear asmcp__<server>__<tool>.
- Skill/command not found: Run
/reload-plugins. Confirm the plugin is enabled with/plugin list. - MCP not attached: Run
/rhombus-mcp-status(developer plugin). Check thatRHOMBUS_API_KEYis exported in the shell that launched Claude Code. - Unexpected behavior: Open the relevant
SKILL.md, command, or agent — the instructions are plain Markdown and readable directly. - Questions about creating skills or commands: See contributing.md.