Security audit

Dependency Map Builder

Security checks across malware telemetry and agentic risk

Overview

This skill is a local dependency-report helper with disclosed Python-based file input/output and no evidence of network access, credential use, persistence, or destructive behavior.

Install this if you want a local dependency-map/report generator. Review the input and output paths before running the Python helper, prefer --dry-run or stdout for sensitive material, and avoid editing the bundled spec to enable unrelated audit modes unless you understand they will inspect local files.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
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

Findings (5)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 90% confidence
Finding: The skill declares no explicit permissions, yet its instructions allow use of python3 with file input/output and shell execution. This creates a capability transparency problem: operators may treat it as a low-risk planning skill while it can access local files and invoke executables, increasing the chance of unintended data exposure or unsafe execution in agent environments.

Description-Behavior Mismatch

High

Confidence: 97% confidence
Finding: The script’s implemented behaviors are broad local auditing and report-generation modes that do not align with the declared dependency-map-builder purpose. This capability mismatch is dangerous because it can cause users to run the skill on arbitrary directories and files under a misleading description, expanding access to local project content and enabling unintended inventorying of repositories.

Context-Inappropriate Capability

Medium

Confidence: 93% confidence
Finding: The code contains built-in pattern scanning for secrets, shell execution strings, and private URLs, which is unrelated to dependency mapping. In this skill context, that hidden extra capability increases data exposure risk because users may provide sensitive repositories expecting coordination output, while the tool inspects and surfaces security-relevant content from local files.

Context-Inappropriate Capability

Medium

Confidence: 94% confidence
Finding: The skill performs package compliance auditing of skill directories, including enumerating expected files and parsing SKILL.md metadata, which is unrelated to the stated cross-team dependency workflow. This broadens the tool’s inspection surface and makes the skill misleading, allowing collection of repository structure and metadata under a benign-seeming project-management label.

Intent-Code Divergence

Low

Confidence: 81% confidence
Finding: The CLI help text describes a generic local support script, which contradicts the specific dependency-map-builder metadata. While not directly exploitable like code execution, this mismatch is a security-relevant trust issue because vague labeling can conceal broader file-analysis behavior and mislead users about what the skill will do with their local inputs.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.