Security audit

AI懂帮我巧匠

Security checks across malware telemetry and agentic risk

Overview

This skill is mostly an onboarding/profile generator, but it under-discloses persistent storage of personal memories and explicitly encourages a plaintext secrets file.

Only install this if you are comfortable with the agent creating local profile, memory, and document files about you. Do not put passwords, tokens, financial identifiers, medical details, or other secrets into SECRET.md or similar markdown files; use a real secret manager instead. Review and delete generated files when they are no longer needed.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Intent-Code Divergence

High

Confidence: 98% confidence
Finding: The skill repeatedly assures users that data is not stored and only exists on their device, while elsewhere instructing creation of persistent files containing personal and potentially sensitive information. This mismatch can mislead users into disclosing more data than they otherwise would, undermining informed consent and increasing privacy risk.

Context-Inappropriate Capability

High

Confidence: 99% confidence
Finding: Directing the agent to create a dedicated SECRET.md file encourages centralized retention of confidential information without a clear necessity for the skill's stated purpose. Consolidating secrets into a plaintext document materially increases the blast radius if the device, workspace, or downstream tooling is compromised.

Ssd 3

High

Confidence: 98% confidence
Finding: The skill explicitly frames long-term retention of user details, memory, tool history, and secret information as part of its design. Persistently compiling this information increases the likelihood of sensitive data accumulation, unauthorized disclosure, and misuse far beyond the minimum needed to generate onboarding documents.

Ssd 3

High

Confidence: 99% confidence
Finding: Instructing creation of a SECRET.md file for repeatedly used confidential information normalizes insecure secret retention in a human-readable file. Such files are easily copied, indexed, synced, or exposed to other tools, making compromise more likely and more damaging.

Ssd 3

High

Confidence: 98% confidence
Finding: The templates instruct the agent to record user profile data, feedback history, long-term memory, tool-use history, and sensitive information for future use. This creates a durable dossier that can exceed user expectations, increases secondary-use risk, and becomes especially dangerous because the skill simultaneously reassures users that nothing is stored.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.