Security audit

Port Manager

Security checks across malware telemetry and agentic risk

Overview

This is a local port-management helper that clearly discloses port inspection and process termination, but users should be careful because freeing a port stops the owning process.

Install only if you want a local tool for resolving port conflicts. Use query or list before free/check, confirm termination only for processes you recognize, and consider clearing the bundled ports.json so records start with your own environment.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (5)

Context-Inappropriate Capability

Medium

Confidence: 92% confidence
Finding: The free and check commands can terminate any local process bound to a user-supplied port, with only a simple interactive confirmation and no validation that the process belongs to the expected service or was previously recorded by this tool. In an agent or automation context, this creates a denial-of-service risk because an operator could inadvertently kill unrelated or important local services by targeting a commonly used port.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill instructs users to free an occupied port by terminating the process, but it does not clearly warn that this may kill an unrelated or critical service. In a port-management skill, users are likely to invoke this during troubleshooting, which increases the chance of accidental service disruption if the process identity and consequences are not surfaced first.

Missing User Warnings

Medium

Confidence: 98% confidence
Finding: The implementation example includes `kill $(lsof -t -i :5432)` without any user-facing safety checks, process verification, or recommendation for graceful termination. This is dangerous because it normalizes blind termination of whatever holds the port, which can cause denial of service or data loss if the process is important.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill instructs users to free an occupied port by terminating the process, but it does not warn that this can abruptly stop legitimate services, cause data loss, or interrupt dependent applications. In an agent-driven context, presenting destructive operational guidance without an explicit confirmation and impact warning increases the chance of unsafe or unintended process termination.

Missing User Warnings

Medium

Confidence: 98% confidence
Finding: The implementation example shows `kill $(lsof -t -i :5432)` with no safeguards, verification step, or warning about consequences. This is dangerous because it normalizes immediate termination of whatever process is listening on the port, which could include production services or critical local software, especially if followed by an automated agent or copied verbatim by a user.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal