OpenClaw · Skills: Media & design

Skills from verified OpenClaw publishers, and official plugins. Each entry links to its ClawHub page, which shows the current security scan.

NVIDIA TAO skill for CLIP: retrieval, classification, embeddings, and TensorRT deployment.

NVIDIA TAO skill for Cosmos-Embed1 video embeddings, search, and fine-tuning.

NVIDIA TAO skill for the shared Cosmos3 frontend for Cosmos-RL and framework training.

NVIDIA TAO skill for fine-tuning a Hugging Face model when no TAO model skill fits.

NVIDIA TAO skill for InternVideo2-CLIP video retrieval, LoRA tuning, and deployment.

NVIDIA TAO skill for converting a parquet of image paths into embeddings.

NVIDIA TAO skill for a two-step pipeline that grounds referring expressions to bounding boxes.

NVIDIA TAO skill for creating referring expressions from images and KITTI boxes.

NVIDIA TAO skill for producing reasoning training data from raw video.

NVIDIA TAO gate that must run before launching any TAO workflow on a platform.

NVIDIA TAO skill for embedding and mining source images for VCN AOI iterations.

NVIDIA TAO skill for AutoML and hyperparameter search on TAO networks in containers.

NVIDIA TAO skill for the end-to-end DEFT improvement loop on PCB inspection models.

NVIDIA TAO skill for the disk-based DEFT AOI loop on Cosmos Reason 3.

NVIDIA TAO skill for mining-driven DEFT improvement on binary ITS video questions.

NVIDIA TAO skill for iteratively improving CLIP or SigLIP retrieval on attribute data.

NVIDIA TAO skill for starting, querying, and stopping a network-specific inference microservice.

NVIDIA TAO skill for training video action recognition on RGB, optical flow, or joint inputs.

NVIDIA TAO skill for training monocular depth estimation with Depth Anything v2.

NVIDIA TAO skill for training FastFoundationStereo real-time stereo depth models.