Security audit

代码解释工具免费版

Security checks across malware telemetry and agentic risk

Overview

This code-explanation skill appears non-malicious, but it asks for broad command execution and file-write authority that is not clearly needed or well scoped for explaining code.

Review this skill before installing. It is suitable in concept for explaining code, but only install it if you are comfortable granting shell execution and file-write capability to a broadly triggered code-help skill. Prefer using it in a limited workspace and avoid pointing it at sensitive proprietary code unless you understand how your Agent's LLM processes that code.

SkillSpector

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Behavioral ASTexec() Call, eval() Call, Dynamic Import
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands

Findings (5)

Context-Inappropriate Capability

Medium

Confidence: 94% confidence
Finding: The skill is presented as a code-explanation helper, but it declares write capability without a clear, necessary feature that requires modifying files. In an agent environment, unnecessary write access expands the attack surface and could let ordinary explanation prompts be escalated into file creation or modification actions if the agent misinterprets instructions or follows prompt-injected code comments.

Context-Inappropriate Capability

Medium

Confidence: 96% confidence
Finding: The skill declares exec despite describing a local code-understanding workflow that mainly needs read-only access. Command execution is materially more dangerous than explanation functionality requires, because it can run arbitrary shell commands against the host environment if triggered directly or through prompt injection in analyzed code or user input.

Intent-Code Divergence

Medium

Confidence: 87% confidence
Finding: The documentation claims that all code analysis is completed locally, yet the dependency section also states that an LLM API is required and error handling references network-sensitive behavior. This can mislead users into sharing proprietary code under the false assumption that no remote processing or transmission occurs, creating confidentiality and trust risks.

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The activation phrases are extremely broad and overlap with normal user conversation, making accidental invocation plausible. In a skill with read/exec/write-capable tooling, unintended activation increases the chance that routine discussion about code causes file access or other side effects without the user's informed intent.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill advertises exec and write capabilities and mentions create/query/export operations, but does not clearly warn users that it may read local files or perform filesystem-affecting actions. This lack of transparency undermines informed consent and is especially risky because users may think they are invoking a harmless explainer while the skill has powerful host-level capabilities.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.