cdf-image-attribute-augmentation — Claude Code skill
cdf-image-attribute-augmentation — a skill that automates an image-attribute augmentation pipeline: prepares person-crop data, synthesizes clothing variants via Cosmos, retrieves attributes, and generates annotated datasets on Kubernetes.
no license ★ 4,106 updated 2026-10-23
npx skills add NVIDIA/skills --skill cdf-image-attribute-augmentation -a claude-codeWhat cdf-image-attribute-augmentation does
The skill launches the end-to-end DAG image_attribute_augmentation_dag_k8s on Kubernetes. You provide a folder of source crop images (structured by person ID → files), and the skill constructs a payload, verifies Airflow controller and service readiness, then triggers the run via API. The DAG chains preprocessing, clothing synthesis in Cosmos, post-processing, attribute extraction for persons and events, and result storage to S3.
You choose between external endpoints (VLM, LLM, image-edit services — provide their HTTPS URLs) or internal mode, where the DAG deploys services within the cluster. Input and output S3 paths are required; optional parameters include the number of person folders to process, clothing variants per person, and an attribute distribution file. The skill requires cluster credentials (you supply the config file path) and validates the Kubernetes manifest before triggering.
What it helps with
- Synthesize clothing variants for a person re-identification dataset
- Generate annotated crop images with extracted attributes stored in S3
- Run an augmentation pipeline without manual Airflow setup
Example prompts for the agent
- Run augmentation on the folder
s3://bucket/people/with 100 crop images, save results tos3://bucket/augmented/ - Trigger the DAG with 5 clothing variants per person and a custom attribute distribution from
clothing-dist.yaml - Check the status of a running pipeline and fetch the link to the annotated output dataset
How to install cdf-image-attribute-augmentation
1Skills CLI (Vercel)
npx skills add NVIDIA/skills --skill cdf-image-attribute-augmentation -a claude-code2Manually: put the files in place
~/.claude/skills/cdf-image-attribute-augmentation/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/cdf-image-attribute-augmentation/
1Skills CLI (Vercel)
npx skills add NVIDIA/skills --skill cdf-image-attribute-augmentation -a codex2Manually: put the files in place
.agents/skills/cdf-image-attribute-augmentation/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add NVIDIA/skills --skill cdf-image-attribute-augmentation -a cursor2Manually: put the files in place
.cursor/skills/cdf-image-attribute-augmentation/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install NVIDIA/skills skills/cdf-image-attribute-augmentation2Manually: put the files in place
.github/skills/cdf-image-attribute-augmentation/SKILL.md1Skills CLI (Vercel)
npx skills add NVIDIA/skills --skill cdf-image-attribute-augmentation -a gemini-cli2Platform CLI
gemini skills install https://github.com/NVIDIA/skills --path skills/cdf-image-attribute-augmentationManually: put the files in place
.devin/skills/cdf-image-attribute-augmentation/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about cdf-image-attribute-augmentation
How do I install cdf-image-attribute-augmentation in Claude Code?
Run npx skills add NVIDIA/skills --skill cdf-image-attribute-augmentation -a claude-code, then open ~/.claude/skills/cdf-image-attribute-augmentation/SKILL.md to verify installation.
What does cdf-image-attribute-augmentation do?
It triggers a Kubernetes DAG that expands your image dataset: generates clothing variants for person crops via Cosmos, extracts attributes (events, person traits), and saves the annotated result to S3.
What inputs does the skill need?
An S3 or HTTPS path with folders per person ID and images, an S3 output path, and a choice of mode (external service endpoints or in-cluster services). For external endpoints, provide HTTPS URLs for the VLM, LLM, and image-edit services.