Security audit

Convert Code Snippets

Security checks across malware telemetry and agentic risk

Overview

This code-conversion skill is plausible, but it asks for an unspecified API key and documents a write mode without explaining what data may be changed or sent elsewhere.

Review before installing. Do not provide an API key or use this with proprietary code, secrets, or internal schemas until the publisher documents what service uses the key, what data is sent externally, and exactly how write mode handles files and overwrites.

SkillSpector

By NVIDIA

Vulnerability Patterns

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
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (6)

Intent-Code Divergence

Medium

Confidence: 93% confidence
Finding: The README presents generic read/write operating modes that do not match the declared purpose of a code-conversion skill. This can mislead users or downstream agents into invoking the skill with broader file-access or data-modification expectations, increasing the chance of unintended writes or unsafe delegation.

Intent-Code Divergence

Medium

Confidence: 88% confidence
Finding: Claiming that an API key is required for a local code-conversion skill introduces unnecessary credential expectations and may prompt users to disclose secrets where none are needed. In agent ecosystems, misleading credential requirements are risky because they normalize secret collection and can facilitate credential harvesting or overprivileged deployments.

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The README leaves the trigger condition as a placeholder ('When triggered: [trigger condition]') instead of defining concrete activation criteria. In an agent skill, ambiguous triggering can cause the skill to run in unintended contexts, increasing the chance of inappropriate code transformation, mishandling sensitive snippets, or bypassing safer specialized workflows.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The README advertises a '--mode write' operation with an input file but provides no warning that this mode may modify files or produce side effects. Users or calling agents may assume the skill is read-only and invoke write mode on important data, leading to accidental overwrites, corruption, or unsafe automation when handling code and structured inputs.

Vague Triggers

Medium

Confidence: 82% confidence
Finding: A placeholder trigger condition leaves activation criteria undefined, which is dangerous in an agent skill because ambiguous routing can cause the tool to run in unintended contexts. When combined with misleading documentation elsewhere in the file, this ambiguity increases the risk of over-broad invocation and unexpected behavior.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The usage examples advertise a write mode with an input file but provide no warning about data modification, overwrite behavior, or output destination. In a skill ecosystem, undocumented write capabilities are security-relevant because users and agents may assume the tool is read-only and inadvertently permit file changes.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.