deepstream-generate-pipeline — Claude Code skill
Builds gst-launch-1.0 commands for DeepStream from a described task. Use it for video or image inference, tracking, streaming, and output handling.
npx skills add NVIDIA/skills --skill deepstream-generate-pipeline -a claude-codeWhat deepstream-generate-pipeline does
The skill asks for the pipeline source, stream count, inference mode, tracker, sink, platform, and transforms. It then retrieves a ready-to-run gst-launch-1.0 pipeline with BM25 search across more than 270 verified DeepStream pipelines.
Its files include scripts/generate_pipeline.py for retrieval and scripts/validate_pipeline.py for syntax, element, property, and live-parse checks. It covers local video and images, RTSP streams, USB cameras, test patterns, nvinfer, nvinferserver, plus NvDCF, IOU, NvSORT, and DeepSORT trackers.
You need Python 3.8+, DeepStream SDK at /opt/nvidia/deepstream/deepstream/, and gst-launch-1.0 with gst-inspect-1.0 on PATH. It requires an x86 dGPU or aarch64 with Jetson or SBSA, so check the target environment first.
What it helps with
- build a pipeline for local-video inference
- configure detection and tracking for RTSP streams
- save processed output as MP4, H264, or images
- apply rotation, cropping, or resizing
- validate GStreamer elements and properties
Example prompts for the agent
- Build a pipeline to run inference on an image.
- Create a detection pipeline with NvDCF tracking for 4 RTSP streams and display the output.
- Generate a
gst-launch-1.0pipeline that processes a video and saves the result as MP4.
How to install deepstream-generate-pipeline
1Skills CLI (Vercel)
npx skills add NVIDIA/skills --skill deepstream-generate-pipeline -a claude-code2Manually: put the files in place
~/.claude/skills/deepstream-generate-pipeline/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/deepstream-generate-pipeline/
1Skills CLI (Vercel)
npx skills add NVIDIA/skills --skill deepstream-generate-pipeline -a codex2Manually: put the files in place
.agents/skills/deepstream-generate-pipeline/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add NVIDIA/skills --skill deepstream-generate-pipeline -a cursor2Manually: put the files in place
.cursor/skills/deepstream-generate-pipeline/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install NVIDIA/skills deepstream-generate-pipeline2Manually: put the files in place
.github/skills/deepstream-generate-pipeline/SKILL.md1Skills CLI (Vercel)
npx skills add NVIDIA/skills --skill deepstream-generate-pipeline -a gemini-cli2Platform CLI
gemini skills install https://github.com/NVIDIA/skills --path skills/deepstream-generate-pipelineManually: put the files in place
.devin/skills/deepstream-generate-pipeline/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add NVIDIA/skills --skill deepstream-generate-pipeline -a opencodeClaude.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 deepstream-generate-pipeline
How do I install deepstream-generate-pipeline in Claude Code?
Run npx skills add NVIDIA/skills --skill deepstream-generate-pipeline -a claude-code. The skill is placed at ~/.claude/skills/deepstream-generate-pipeline/SKILL.md.
What does deepstream-generate-pipeline do?
It gathers DeepStream pipeline requirements and produces a gst-launch-1.0 command. A local BM25 retriever selects a pipeline from verified examples.
What is required to run the skill?
You need Python 3.8+, DeepStream SDK, plus gst-launch-1.0 and gst-inspect-1.0 on PATH. The environment must use an x86 dGPU or aarch64 with Jetson or SBSA.
