Security audit

JSON Schema Validator / JSON 架构验证器

Security checks across malware telemetry and agentic risk

Overview

This is a lightweight JSON/YAML validation skill with no bundled executable or install payload, but users should treat its auto-fix option as write-capable.

Installers should understand that this skill may guide an agent to modify JSON/YAML files when --auto-fix is used. Prefer version-controlled files, backups, or a read-only validation run before applying fixes, and note that the declared curl dependency is not explained by the current artifact.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The skill prominently advertises an auto-fix capability but does not warn that using it may modify user files. That creates a real safety issue because users may invoke the tool expecting read-only validation and unintentionally overwrite configuration or data files, potentially causing data loss or introducing unnoticed changes into deployments.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The example command includes --auto-fix with no caution about overwriting or altering the target file, which normalizes destructive behavior as a default copy-paste action. In a skill context, examples are often executed verbatim, so omission of a warning materially increases the chance of unintended file modification and silent corruption of important config artifacts.

VirusTotal

62/62 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.