Security audit

source-driven-develo

Security checks across malware telemetry and agentic risk

Overview

This skill is broadly framed and requests command execution despite unclear scope, vague triggers, and mixed claims about documentation guidance versus automation.

Review this before installing. It is not clearly malicious and contains no executable installer, but it asks for command execution and discusses credentials, external APIs, file handling, and automation without tight scope. Only use it in a controlled workspace, require explicit approval before commands or external connections, and avoid giving it secrets unless the publisher clarifies how they are handled.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands

Findings (6)

Description-Behavior Mismatch

High

Confidence: 97% confidence
Finding: The manifest markets the skill as documentation-grounded, source-cited development guidance, but the body instead describes generic automation, data processing, and command-execution behavior. This mismatch can mislead users and orchestration systems into invoking a materially more powerful skill than expected, increasing the chance of unsafe execution and policy bypass.

Context-Inappropriate Capability

Medium

Confidence: 93% confidence
Finding: The skill declares an exec capability even though its stated purpose is documentation-grounded development assistance, which does not inherently require shell or process execution. Unnecessary execution privileges violate least privilege and enlarge the attack surface if user-controlled input is later routed into commands or scripts.

Intent-Code Divergence

High

Confidence: 96% confidence
Finding: The core-function and feature sections contradict the claimed documentation-grounding purpose and instead present a broad workflow engine for data handling and orchestration. In an agent ecosystem, this kind of capability drift is dangerous because reviewers and users may authorize the skill under a safer mental model than what the content actually enables.

Vague Triggers

High

Confidence: 90% confidence
Finding: The trigger condition 'when the user needs related operations' is so broad that the skill may be invoked in many contexts without meaningful intent matching. For a skill with read/exec capability and references to APIs, files, and commands, overbroad triggering increases the risk of unexpected activation and unsafe actions.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The description does not clearly warn users that the skill includes command execution capability. Users expecting only documentation-grounded advice may unknowingly authorize a skill that can run commands, creating consent and safety issues, especially in automated agent routing contexts.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The setup instructions mention API keys, credentials, and external connections, but the skill lacks a clear, up-front warning about handling secrets and communicating with external services. This can lead to accidental credential exposure, unsafe secret storage, or unanticipated data egress by users who believe the skill is only local documentation assistance.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.