dominodatalab — Claude Code plugin: what it does, how to install
A Claude plugin that handles Domino Data Lab work: workspaces, jobs, environments, datasets and app deployment.
/plugin install dominodatalab@claude-plugins-official
/plugin marketplace add anthropics/claude-plugins-officialWhat dominodatalab does
The plugin brings Domino Data Lab operations into a Claude session. You get help configuring Jupyter, VS Code and RStudio workspaces, running batch and scheduled jobs, building custom Docker environments, versioning datasets with snapshots, and deploying React, Streamlit or Dash apps. Model endpoints, Spark, Ray and Dask clusters, and agent tracing through the Domino SDK are covered too.
Skills are auto-invoked from task context, and slash commands such as /domino-app-init are available. A bundled Domino MCP server requires the uv package manager. You also need access to a Domino instance and a Domino API key when working outside a Domino workspace; Claude CLI must be 1.0.33 or later.
One thing to know before installing: the domino-claude-plugin repository is deprecated and gets no new releases. Development moved to dominodatalab/domino-ai-plugin, which is where issues and pull requests go now. An existing clone of the old repo keeps loading frozen content until you point its origin at the new one.
What it helps with
- configure a Jupyter or RStudio workspace for a specific task
- launch a batch job and check its status
- build a custom Docker environment with the packages you need
- snapshot a dataset and share it with the team
- deploy a Streamlit app or a model endpoint
Example prompts for the agent
- Help me deploy a Streamlit app to Domino
- Create a dataset snapshot and share it with my team
- Spin up a Ray cluster for distributed training
How to install dominodatalab
Plugin marketplace
/plugin install dominodatalab@claude-plugins-official/plugin marketplace add anthropics/claude-plugins-officialStep by step, with commands for each platform: Claude Code plugins: marketplaces, installing, and building your own
Questions about dominodatalab
How do I install dominodatalab in Claude Code?
From the official marketplace: run /plugin marketplace add anthropics/claude-plugins-official, then /plugin install dominodatalab@claude-plugins-official.
What does the dominodatalab plugin do?
It gives Claude access to Domino Data Lab platform tasks: workspaces, jobs, environments, datasets, app and model deployment, distributed computing and agent tracing.
What do I need before installing it?
Claude CLI 1.0.33 or later, access to a Domino instance, a Domino API key for operations outside a workspace, and the uv package manager for the bundled MCP server.
Source text
SKILL.md · MIT
# This repository has moved to [dominodatalab/domino-ai-plugin](https://github.com/dominodatalab/domino-ai-plugin)
> [!CAUTION]
> **`domino-claude-plugin` is deprecated and receives no new releases.** The Domino plugin now
> lives at **[github.com/dominodatalab/domino-ai-plugin](https://github.com/dominodatalab/domino-ai-plugin)**,
> for Claude Code, Codex, Cursor, Gemini CLI and Antigravity. Install it, report issues and open
> pull requests there. Issues and pull requests are closed here.
>
> **Switch an existing clone to the new repository** and start a new session:
>
> ```bash
> P=~/.claude/marketplaces/domino/plugins/domino-claude-plugin # or the path of your clone
> git -C $P remote set-url origin https://github.com/dominodatalab/domino-ai-plugin.git
> git -C $P fetch origin main
> git -C $P checkout -B main origin/main
> ```
>
> Every other install method: [Installation](https://github.com/dominodatalab/domino-ai-plugin#installation-claude-code).
---
# Domino Data Lab Plugin for Claude Code
A comprehensive Claude Code plugin providing full coverage of the Domino Data Lab platform for AI-assisted development.
## Overview
This plugin enables Claude Code to help you with all aspects of Domino Data Lab, including:
- **Workspaces**: Jupyter, VS Code, RStudio configuration and management
- **Jobs**: Batch execution, scheduled jobs, and monitoring
- **Environments**: Custom Docker environments and package management
- **Datasets**: Data versioning with snapshots and sharing
- **NetApp Volumes**: Enterprise-grade multi-terabyte storage with near-instant snapshots
- **Apps**: Deploy React, Streamlit, and Dash applications
- **Models**: Deploy, monitor, and manage model endpoints
- **GenAI**: Trace and evaluate AI agents with the Domino SDK
- **Distributed Computing**: Spark, Ray, and Dask clusters
- **And more...**
## Installation
### Prerequisites
- **Claude Code CLI** v1.0.33 or later (`claude --version` to check)
- Access to a Domino Data Lab instance
- Domino API key (for API operations when running outside a Domino workspace)
- **`uv`** package manager ([install guide](https://github.com/astral-sh/uv)) — required for the bundled Domino MCP server
---
### Option 1: Marketplace Install (Recommended)
This approach registers the plugin through Claude Code's native marketplace system so it persists across sessions.
**Step 1: Clone the repository and create a marketplace wrapper**
```bash
# Clone the plugin
git clone https://github.com/dominodatalab/domino-claude-plugin.git
# Create the marketplace directory structure
mkdir -p ~/.claude/marketplaces/domino/.claude-plugin
mkdir -p ~/.claude/marketplaces/domino/plugins
# Move the plugin into the marketplace
mv domino-claude-plugin ~/.claude/marketplaces/domino/plugins/domino-claude-plugin
```
**Step 2: Create the marketplace manifest**
```bash
cat > ~/.claude/marketplaces/domino/.claude-plugin/marketplace.json << 'EOF'
{
"name": "domino-marketplace",
"owner": {
"name": "Domino Data Lab",
"email": "[email protected]"
},
"plugins": [
{
"name": "domino-claude-plugin",
"description": "Domino Data Lab plugin for Claude Code - workspaces, jobs, environments, datasets, apps, models, and more",
"source": "./plugins/domino-claude-plugin",
"category": "development"
}
]
}
EOF
```
**Step 3: Register the marketplace and install the plugin**
Launch Claude Code and run:
```
/plugin marketplace add /home/<your-username>/.claude/marketplaces/domino
/plugin install domino-claude-plugin@domino-marketplace
```
> **Note:** Replace `<your-username>` with your actual username, or use the full absolute path (e.g., `/home/ubuntu/.claude/marketplaces/domino`). The `~` shorthand may not expand correctly.
**Step 4: Restart Claude Code**
```
/exit
claude
```
The plugin should appear in your loaded plugins on startup. Verify with:
```
/plugin
```
Navigate to the **Installed** tab to confirm `domino-claude-plugin` is listed.
---
### Option 2: Direct Plugin Directory (Development / Quick Start)
Use the `--plugin-dir` flag to load the plugin directly. This is ideal for development, testing, or quick evaluation.
```bash
# Clone the plugin (the manifest .claude-plugin/plugin.json is part of the repo; do not overwrite it)
git clone https://github.com/dominodatalab/domino-claude-plugin.git
# Run Claude Code with the plugin loaded in place
claude --plugin-dir ./domino-claude-plugin
```
Loaded this way the plugin is read in place: a `git pull` (or checking out a `release-*` branch
or tag) takes effect at the next session with no reinstall.
To make this persistent without the marketplace approach, add a shell alias:
```bash
echo 'alias claude="claude --plugin-dir /path/to/domino-claude-plugin"' >> ~/.bashrc
source ~/.bashrc
```
### Versions and updates
The plugin version is `YYYY.DDD.N` (for example `2026.603.3`): the release year, the Domino
line the content is correct for as a floor (`603` = Domino 6.3 and later, including Cloud; the
major followed by the two-digit minor), and a release counter. Each release is tagged
`release-YYYY.DDD.N` and published as a GitHub Release. Marketplace installs (Option 1
and the Anthropic marketplace) are cached by that version string and update only when it
changes; check what you have with `claude plugin list` and update with
`claude plugin update <name>@<marketplace>`. See CONTRIBUTING.md "Release branches, tags and
backports" for the branch and tag policy.
---
### Option 3: Team / Project-Level Install
For teams sharing a project, add the marketplace to your project's `.claude/settings.json`:
```json
{
"extraKnownMarketplaces": {
"domino-marketplace": {
"source": {
"source": "directory",
"path": "/path/to/domino-marketplace"
}
}
},
"enabledPlugins": {
"domino-claude-plugin@domino-marketplace": true
}
}
```
When team members trust the repository folder, Claude Code will prompt them to install the marketplace and plugin automatically.
---
## Verifying the Installation
After installation, test that the plugin is working:
1. **Check slash commands are available:**
```
/domino-app-init
```
2. **Test skill auto-invocation** by asking a Domino-related question:
```
Help me deploy a Streamlit app to Domino
```
Claude should automatically invoke the `domino-app-deployment` skill.
3. **Check the plugin is listed:**
```
/plugin
```
Navigate to the **Installed** tab.
> **Note:** Plugin skills do not appear in the `/skills` list. They are auto-invoked by Claude based on task context and will show in Claude's init message at the top of a new session.
---
## Updating the Plugin
How you update depends on how you installed:
- **Marketplace install (Option 1, Option 3, or the Anthropic marketplace):** Claude Code keeps a
cached copy keyed by the plugin version. Pulling the source directory does **not** change what
Claude loads. Run the update command and restart:
```bash
claude plugin update domino-claude-plugin@domino-marketplace # local marketplace (Option 1 / DSE)
claude plugin update dominodatalab@claude-plugins-official # Anthropic marketplace
```
If it reports "already at the latest version", the published version has not changed yet
(see "Versions and updates" above).
- **`--plugin-dir` (Option 2):** the plugin is read in place. `git pull` in the clone, or check
out a `release-*` branch or tag, then start a new session.
---
## What's Included
### Bundled MCP Server
The plugin includes a vendored copy of the [Domino MCP Server](https://github.com/dominodatalab/domino_mcp_server) that **starts automatically** when the plugin is enabled. It provides tools for running Domino jobs, checking job status/results, and syncing files with DFS-based projects.
- **Inside a Domino workspace:** Fully automatic — authentication uses ephemeral tokens, project info is auto-detected.
- **Outside Domino (laptop):** Set `DOMINO_API_KEY` and `DOMINO_HOST` as environment variables in your shell.
Requires `uv` to be installed (see [Prerequisites](#prerequisites)).
### Skills (20 Total)
| Skill | Description |
| --- | --- |
| `domino-workspaces` | Jupyter, VS Code, RStudio workspace management |
| `domino-jobs` | Jobs and scheduled jobs execution |
| `domino-environments` | Compute environments and Dockerfile customization |
| `domino-datasets` | Data management, snapshots, and versioning |
| `domino-projects` | Git integration and project collaboration |
| `domino-app-deployment` | Deploy web apps (React, Streamlit, Dash) |
| `domino-experiment-tracking` | MLflow experiment tracking and model registry |
| `domino-genai-tracing` | `@add_tracing` decorator and `DominoRun` |
| `domino-model-endpoints` | Deploy and call model APIs |
| `domino-model-monitoring` | Drift detection and model quality tracking |
| `domino-flows` | Flyte-based workflow orchestration |
| `domino-distributed-computing` | Spark, Ray, Dask cluster management |
| `domino-ai-gateway` | LLM proxy for OpenAI, Bedrock, etc. |
| `domino-launchers` | Parameterized web forms for self-service |
| `domino-modeling-assistant` | MCP server for AI-assisted model development |
| `domino-data-connectivity` | S3 Mountpoint, AWS IRSA, Azure credentials |
| `domino-python-sdk` | Python SDK (python-domino) and REST API |
| `domino-data-sdk` | Data SDK (domino-data) for data sources, datasets, training sets |
| `domino-ui-design` | Knowledge on Domino UI styling for integrated App design |
| `netapp-volumes` | Enterprise-grade multi-terabyte NetApp ONTAP storage with near-instant snapshots and versioning |
### Slash Commands
| Command | Description |
| --- | --- |
| `/domino-app-init` | Initialize a new Domino app with framework templates |
| `/domino-debug-proxy` | Debug reverse proxy issues for apps |
| `/domino-experiment-setup` | Set up MLflow experiment tracking |
| `/domino-trace-setup` | Set up GenAI tracing with the Domino SDK |
### Subagents
| Agent | Description |
| --- | --- |
| `domino-deploy` | Specialized agent for deploying apps, models, and endpoints |
| `domino-debug` | Agent for debugging Domino issues and troubleshooting |
| `domino-setup` | Agent for setting up new projects and configurations |
### Output Styles
Switch output styles with `/output-style`:
| Style | Description |
| --- | --- |
| `domino-learning` | Educational mode with Domino Insights after each task |
| `domino-mlops` | Production-focused with MLOps checklists and best practices |
---
## Project Structure
```
domino-claude-plugin/
├── .claude-plugin/
│ └── plugin.json # Plugin manifest
├── .mcp.json # Bundled MCP server config (auto-starts)
├── mcp-servers/ # Vendored MCP servers
│ └── domino_mcp_server/ # Domino MCP Server (jobs, DFS sync)
├── agents/ # Subagents
│ ├── domino-deploy.md
│ ├── domino-debug.md
│ └── domino-setup.md
├── output-styles/ # Custom output styles
│ ├── domino-learning.md
│ └── domino-mlops.md
├── skills/ # 18 skill directories
│ ├── workspaces/
│ ├── jobs/
│ ├── environments/
│ ├── datasets/
│ ├── projects/
│ ├── app-deployment/
│ ├── experiment-tracking/
│ ├── genai-tracing/
│ ├── model-endpoints/
│ ├── model-monitoring/
│ ├── flows/
│ ├── distributed-computing/
│ ├── ai-gateway/
│ ├── launchers/
│ ├── modeling-assistant/
│ ├── data-connectivity/
│ ├── python-sdk/
│ ├── domino-data-sdk/
│ └── netapp-volumes/
├── commands/ # Slash commands
├── hooks/ # Example automation hooks
├── templates/ # Code templates
│ ├── vite-react/
│ ├── streamlit/
│ ├── dash/
│ ├── experiment/
│ └── tracing/
├── CONTRIBUTING.md
├── LICENSE
└── README.md
```
---
## Troubleshooting
| Issue | Solution |
| --- | --- |
| `/skills` shows "No skills found" | Plugin skills don't appear in `/skills` — they are auto-invoked by Claude based on context. Check `/plugin` → Installed tab instead. |
| Plugin not loading from settings.json | Claude Code does **not** support a `"plugins"` array in `settings.json`. Use the marketplace approach or `--plugin-dir` flag. |
| `~` path not expanding | Always use absolute paths (e.g., `/home/ubuntu/...`) in marketplace commands and settings. |
| Slash commands not appearing | Restart Claude Code after installing. Commands are loaded at session start. |
| "Failed to parse marketplace file" | Ensure `marketplace.json` has the `owner` object and `source` is a string path (e.g., `"./plugins/domino-claude-plugin"`), not a nested object. |
---
## Usage Examples
### Deploy a Streamlit App
```
User: Help me deploy a Streamlit dashboard to Domino
Claude: I'll help you set up a Streamlit app for Domino...
```
### Set Up Experiment Tracking
```
User: /domino-experiment-setup
Claude: I'll configure MLflow experiment tracking for your project...
```
### Create a Scheduled Job
```
User: How do I run a training script every day at midnight?
Claude: I'll show you how to create a scheduled job in Domino...
```
### Deploy a Model API
```
User: I need to deploy my scikit-learn model as an API
Claude: I'll help you create a model endpoint in Domino...
```
---
## API Reference
The `domino-python-sdk` skill includes comprehensive REST API documentation:
- `API-PROJECTS.md` — Projects, collaborators, Git repos
- `API-JOBS.md` — Jobs, logs, scheduled execution
- `API-DATASETS.md` — Datasets, snapshots, permissions
- `API-MODELS.md` — Model APIs, deployments, registry
- `API-ENVIRONMENTS.md` — Environments, revisions
- `API-APPS.md` — Apps, versions, instances
- `API-ADMIN.md` — Users, orgs, hardware tiers
---
## Documentation
- [Domino Documentation](https://docs.dominodatalab.com/en/cloud/user_guide/71a047/what-is-domino/)
- [Domino API Guide](https://docs.dominodatalab.com/en/latest/api_guide/f35c19/api-guide/)
- [Domino Blueprints](https://domino.ai/resources/blueprints)
- [python-domino GitHub](https://github.com/dominodatalab/python-domino)
- [Claude Code Plugin Docs](https://code.claude.com/docs/en/plugins)
- [Claude Code Marketplace Docs](https://code.claude.com/docs/en/plugin-marketplaces)
## Contributing
Contributions are welcome! Please feel free to submit issues and pull requests.
## License
This project is licensed under the MIT License — see the [LICENSE](LICENSE) file for details.
## Support
- For Domino platform issues: [Domino Support](https://support.dominodatalab.com/)
- For plugin issues: [GitHub Issues](https://github.com/dominodatalab/domino-claude-plugin/issues)
© Domino Data Lab · MIT · dominodatalab/domino-claude-plugin