Live scan flow powered by NVIDIA SkillSpector

Turn an agent skill into a risk report in seconds.

AgentGateX runs a real static scan, normalizes SARIF evidence, and renders remediation-ready findings for security review.

View existing findings Product tour

Real static scanNo LLM required

Run an agent skill scan

Start with the verified demo skill, or paste a public GitHub/skill URL and AgentGateX will route it through SkillSpector.

or scan a public target

What runs behind the button

AgentGateX calls the scanner service, runs skillspector scan --format sarif, normalizes SARIF, then renders evidence and remediation in this report.

Your scan report will appear here

The report includes severity, evidence location, pattern ID, impact, and remediation guidance mapped into AgentGateX's risk model.

Step 1

Static analysis

Step 2

SARIF normalize

Step 3

Risk report

Verified real scan

The fallback data shown here was generated by running NVIDIA SkillSpector against a local risky sample skill.

Static by default

The default flow uses --no-llm for faster, deterministic demos without external model credentials.

SARIF-native

Findings are normalized from SARIF into AgentGateX evidence, impact, remediation, and risk score fields.