Security audit

Format Converter Free

Security checks across malware telemetry and agentic risk

Overview

This looks like a local format-conversion skill, but it asks for broad execution authority and includes network callback behavior that is not clearly scoped to that purpose.

Review before installing. Use it only on files you explicitly choose, prefer writing to a new output path, back up important configuration or data files, and do not provide callback URLs or external storage credentials unless you intentionally want network integration. Treat any command execution or dependency installation as requiring normal local-shell caution.

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 (5)

Description-Behavior Mismatch

Medium

Confidence: 90% confidence
Finding: The skill presents itself as a local single-file format converter, but it also documents a callback_url parameter and network-oriented behavior. This expands the data flow beyond local conversion and can enable unintended exfiltration of converted data or metadata to remote endpoints, which is inconsistent with user expectations and the stated free-edition scope.

Context-Inappropriate Capability

Medium

Confidence: 88% confidence
Finding: A local data-format conversion skill has no clear need for network-oriented parameters or troubleshooting steps involving connectivity checks. Introducing callback URLs and ping-based guidance normalizes outbound network activity, which could be abused to contact attacker-controlled hosts or leak sensitive file contents under the guise of conversion workflow.

Context-Inappropriate Capability

Medium

Confidence: 93% confidence
Finding: Declaring exec for a simple file-format converter grants arbitrary command execution far beyond what is needed for read/parse/write transformations. In an agent context, this materially increases the attack surface because natural-language requests could be translated into shell commands that read unrelated files, modify the system, or invoke network tools.

Vague Triggers

Medium

Confidence: 80% confidence
Finding: The skill invites unrestricted natural-language requests and says the agent will automatically identify formats and execute the conversion, without tightly constraining accepted operations. In combination with write and exec capabilities, this broad activation model can cause the agent to perform unintended file writes, command execution, or scope creep beyond mere format conversion.

Missing User Warnings

Medium

Confidence: 82% confidence
Finding: The description emphasizes easy conversion but does not clearly warn that outputs may overwrite files or that metadata such as comments, ordering, duplicate keys, or formatting may be lost during conversion. This can lead users to trust the tool with important configuration or data files and suffer silent integrity loss or destructive writes.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.