vss-manage-video-io-storage
Use to call the VIOS REST API (sensor list, timelines, clip extraction, snapshots, add/delete sensors and streams). Not for VLM inference or search.
Security Assessment
About vss-manage-video-io-storage
This skill drives the VIOS (Video I/O and Storage) REST API and the companion NvStreamer API for NVIDIA's VSS blueprint. It solves the problem of managing camera/sensor inventory, RTSP streams, recordings, snapshots, clip extraction, timelines, and video storage by calling documented REST endpoints directly with curl and jq, rather than navigating a UI. It includes explicit routing rules so the agent picks the right path, for example using the direct VIOS file-storage endpoint for a plain MP4 upload versus the NvStreamer upload-to-RTSP-URL handoff only when a live or synthetic RTSP feed is actually requested.
The skill bundles four reference contracts: the full VIOS REST API reference (sensors, storage, snapshots, clip extraction, WebRTC live/replay, RTSP proxy, recorder, service discovery), the NvStreamer API reference (uploads, snapshots, storage info, filesystem scan), and an integration contract describing how VIOS plugs into peer services (RT-VLM, ELK, Kafka, Redis, SDRC). It also documents a sample-data bootstrap that downloads a fixed NGC bundle of eight demo MP4s, with strict instructions never to substitute a similar-sounding file or fabricate upload responses.
Target users are operators and agents running VSS who need to manage video ingestion and storage. The skill deliberately scopes itself away from VLM inference, semantic search, summarization, and analytics (each delegated to a sibling skill). Security-relevant notes: credentials are read from NGC environment variables and the skill fails with an actionable error if they are missing; delete operations are genuinely destructive but are documented with warnings.
FAQ
What can this skill do?
Call the VIOS REST API to add cameras/sensors and RTSP streams, list sensors, check stream status, get snapshots, download clips, upload video files, and manage recordings and storage; it can also drive NvStreamer for synthetic RTSP feeds.
What are the prerequisites?
An active VSS deployment reachable on `$HOST_IP`, NGC credentials in `$NGC_CLI_API_KEY` and `$NVIDIA_API_KEY` for image pulls, and curl, jq, and Docker available on the caller.
What is this skill NOT for?
It is not for VLM inference or visual Q&A (use vss-ask-video), semantic search or ingesting for search (vss-search-archive), narrative summaries (vss-summarize-video), incident reports (vss-generate-video-report), or reading analytics/incidents/alerts (vss-query-analytics).
When should NvStreamer be used instead of a direct VIOS upload?
Only when the user explicitly needs a live/synthetic RTSP camera feed, asks for NvStreamer, or wants an RTSP URL; a plain MP4 upload should use the direct VIOS storage endpoint, not the NvStreamer-to-RTSP handoff.
Are there destructive operations?
Yes. Delete operations for sensors and streams are supported and are destructive; the documentation explains this, though the benchmark noted the delete warning is buried in blockquotes.
All Files
11 filesInstall vss-manage-video-io-storage
Quick Setup:
- Copy the skill folder to
.claude/skills/ - Claude will automatically detect and use the skill
Repository
nvidia/skills