Regression Analysis

Security checks across malware telemetry and agentic risk

Overview

This is a local Excel regression-analysis skill with documentation inconsistencies, but no evidence of hidden access, exfiltration, persistence, or destructive behavior.

Before installing, confirm the intended aggregation method and expected sheet/model count, and be prepared to provide a template file with -t because the advertised built-in template is not included in the scanned artifact. Treat regression outputs as analysis results that should be validated against sample data before business use.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • 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 (1)

Intent-Code Divergence

Medium
Confidence
68% confidence
Finding
The manifest says the regression is based on daily mean data, while the processing steps state daily aggregation uses sums. This mismatch can cause materially different analytical results and may mislead users into trusting outputs produced under different assumptions than advertised, which is a form of integrity risk in an analysis skill.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal