DataDesigner
🎨 NeMo Data Designer: Generate high-quality synthetic data from scratch or from seed data.
DataDesigner is an agent skill from NVIDIA NeMo that enables AI coding assistants to generate high-quality synthetic datasets through natural language instructions. Developers describe the dataset they need, and the agent handles schema design, validation, and generation using the underlying Python framework. The skill integrates with Claude Code and Codex, allowing agents to create diverse text, structured data, and images with controlled relationships, statistical distributions, and validation rules. It supports multimodal workflows, external tool connections via MCP servers, and production-scale synthetic data generation.
Key Features
Use Cases
- 01Training data generation for machine learning models with controlled distributions
- 02Creating synthetic customer reviews with demographic attributes and realistic product categories
- 03Generating test datasets with validated field relationships for application testing
- 04Building multimodal datasets combining text, images, and structured metadata
- 05Producing privacy-safe synthetic alternatives to sensitive production data
- 06Experimenting with different data schemas through agent-assisted iterative design
Related Skills
View moresuperpowers
An agentic skills framework & software development methodology that works.
skills
Skills for Real Engineers. Straight from my .agents directory.
skills
Public repository for Agent Skills
ponytail
Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.
DataDesigner — FAQ
What is DataDesigner?+
DataDesigner is an agent skill and Python framework from NVIDIA NeMo for generating high-quality synthetic datasets. When used as a skill with Claude Code or Codex, you describe the dataset you want and the agent handles schema design, validation, and data generation automatically.
How do I install the DataDesigner skill?+
Install via skills.sh by running 'npx skills add NVIDIA-NeMo/DataDesigner' in your terminal. After installation, invoke the data-designer skill or simply describe your dataset needs to your coding agent.
Which AI assistants work with DataDesigner?+
The skill is tested and confirmed to work with Claude Code and Codex. The underlying Python library can be used programmatically in any environment.
Do I need API keys to use DataDesigner?+
Yes, you need an API key from at least one model provider: NVIDIA Build API, OpenAI, or OpenRouter. Set the corresponding environment variable (NVIDIA_API_KEY, OPENAI_API_KEY, or OPENROUTER_API_KEY) before generating data.
Is DataDesigner free to use?+
DataDesigner is open-source under the Apache 2.0 license and free to use. However, you'll incur costs from your chosen model provider (NVIDIA Build, OpenAI, OpenRouter) based on their pricing. Note that NVIDIA Build is for evaluation/testing only, not production use.
Can DataDesigner connect to external tools and services?+
Yes, DataDesigner supports connections to external tools through local or remote MCP servers, and it can capture interaction traces from these tool calls during data generation.
How do I install DataDesigner?+
Open the source repository on GitHub and follow its README. DataDesigner is a skill — MCP Agents Market links you directly to the official repo.
Is DataDesigner free?+
DataDesigner is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.