NVIDIA JetPack 7.2.1 introduces PyNvVideoCodec 2.2, the NVIDIA Python library for hardware-accelerated video encoding and decoding on NVIDIA GPUs, marking the first time this support is available on Jetson. PyNvVideoCodec produces and consumes video frames as GPU-resident device memory, exposed through the DLPack protocol and CUDA device buffers, and includes features such as multi-mode frame sampling and ThreadedDecoder, which pre-decodes frames in a background thread to decouple decode latency from inference latency. JetPack 7.1 had previously introduced NVIDIA Video Codec SDK support on Jetson Thor, giving C and C++ developers direct access to NVENC and NVDEC. JetPack 7.2.1 also adds foundational agentic video skills above these SDKs, connecting a developer’s goal to live device discovery, supported configurations, working recipes, execution, measurement, and reproducible evidence. These skills turn prompts like “How many H.264 1080p30 streams can this Jetson run for a low-latency use case?” into repeatable codec workflows that inspect the target, choose a supported Video Codec SDK or PyNvVideoCodec path, generate a tested configuration, execute it, and return measured results with warnings and evidence. The unified jetson-videosdk skill allows developers to invoke each workflow independently or combine them to configure and verify codec stages within a broader pipeline