Back to Skills

vss-setup-behavior-analytics

Use to deploy the vss-behavior-analytics service standalone (entrypoint, config-source, optional calibration). Not for the full warehouse deploy.

2,773stars323forksUpdated 8/3/2026

Security Assessment

Safe(91/100)
Security Score91/100

About vss-setup-behavior-analytics

vss-setup-behavior-analytics is an NVIDIA operational skill for deploying the vss-behavior-analytics service as a standalone container, separate from the full warehouse blueprint stack. It solves the problem of bringing up just the spatial-AI analytics pipeline with a user-chosen entrypoint, configuration source, and optional calibration, so operators can run analytics without spinning up the rest of the warehouse profile. The SKILL.md itself handles routing and prerequisites, deferring the full walkthrough (entrypoint table, config-source options, calibration types, dynamic-update wire contract) to reference documents under references/.

Capability-wise it uses Docker Compose to launch the service (for example bringing up vss-behavior-analytics-base from the analytics compose file), requires a repo checkout with VSS_APPS_DIR pointing at the deploy/docker directory for volume binds, and pulls images using NGC credentials supplied via an environment variable. It documents a rich JSON configuration surface consumed by the service's AppConfig — sensors, coordinate reference systems, messaging blocks (Kafka, Redis Streams, MQTT), incident detection thresholds, proximity and trajectory settings — and supports dynamic config and calibration updates over a broker when one is available. Troubleshooting covers connection-refused, NGC 401/403, and GPU out-of-memory cases.

It targets developers and operators who want to run NVIDIA VSS behavior analytics standalone. It performs container deployment and configuration, so it is an operational tool, but credentials are handled through official NGC environment-variable conventions with no exfiltration, and the referenced configuration is standard service tuning. NVSkills-Eval reported a passing evaluation with a 100% Security score.

FAQ

What does this skill deploy?

Just the vss-behavior-analytics container (the spatial-AI analytics pipeline) as a standalone service via Docker Compose, not the full warehouse blueprint stack.

What are the prerequisites?

A repo checkout with $VSS_APPS_DIR pointing at deploy/docker/, NGC credentials in $NGC_CLI_API_KEY for image pulls, Docker Engine 28.3.3 with Compose plugin v2.39.1+, and optionally a Kafka/Redis/MQTT broker.

Does it need a message broker?

No. The container starts without one, though it will cycle in a Restarting state until a broker is reachable. With a broker, dynamic config and calibration updates over mdx-notification become available.

How is it configured?

Through JSON files consumed by AppConfig, covering sensors, coordinate systems, messaging blocks, incident-detection thresholds (all incident types default disabled), and proximity/trajectory settings; example configs ship in the repo.

What common errors are covered?

Connection-refused (service not running), HTTP 401/403 on NGC pulls (missing or expired NGC key), and container OOM or model load failure (insufficient GPU memory).

All Files

11 files
references/configuration.md3.7 KB
View
evals/standalone_deploy.json1.8 KB
View
skill-card.md4.0 KB
View
BENCHMARK.md3.8 KB
View
references/dynamic-calibration.md13.8 KB
View
skill.oms.sig6.0 KB
View
evals/evals.json0.5 KB
View
references/dynamic-config.md19.0 KB
View
references/deploy-behavior-analytics-service.md13.8 KB
View
SKILL.md7.7 KB
View
references/ngc-api-key-registry-login.md2.0 KB
View

Install vss-setup-behavior-analytics

Download and extract the skill files to your .claude/skills/ directory.

Quick Setup:

  1. Copy the skill folder to .claude/skills/
  2. Claude will automatically detect and use the skill

Repository

nvidia/skills