Security audit

image_to_word

Security checks across malware telemetry and agentic risk

Overview

This skill appears to do what it says: upload a user-selected image to Scnet, poll for conversion, and return a temporary Word download link.

Install only if you are comfortable sending selected images to Scnet for processing and receiving remote temporary download links. Keep SCNET_API_KEY in an environment variable or chmod-600 .env file, leave SCNET_API_BASE pointed at a trusted endpoint, avoid sensitive documents unless approved, and treat returned download URLs as private secrets.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (6)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 93% confidence
Finding: The skill documentation shows capabilities to read local files, access environment variables, invoke Python/shell, and send data over the network, but it declares no explicit permissions. This creates a transparency and governance gap: users or platforms may authorize or auto-run the skill without understanding that local files and secrets could be accessed and data sent externally.

Vague Triggers

Medium

Confidence: 87% confidence
Finding: The activation guidance is broad enough that an assistant may auto-trigger this skill based on casual language, increasing the chance that a local file path is sent to an external OCR service without a clear, fresh confirmation. In this skill's context, unintended invocation is more dangerous because the action involves uploading user-provided local content off-device.

Missing User Warnings

High

Confidence: 97% confidence
Finding: The skill lacks a clear warning that local files are uploaded to a third-party service for processing. Because users may provide sensitive images or documents, omission of this disclosure can lead to unintentional exfiltration of private or regulated data to an external provider.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The documentation describes uploading local files or providing public file URLs to a third-party API for asynchronous processing, but it does not clearly warn users that their content leaves the local environment and is processed and stored on external infrastructure. For a document-conversion skill that may handle sensitive images or PDFs, this omission can lead to unintentional disclosure of confidential data to an external service.

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: The result examples expose that converted documents are returned via remotely hosted MinIO pre-signed links, but the documentation does not warn that output files are stored off-platform and accessible through temporary bearer-style URLs. Users may incorrectly assume results remain private or local, increasing the risk of accidental sharing, link leakage, or mishandling of sensitive converted documents.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The script opens a user-supplied local file and uploads it to a third-party API service as part of the conversion workflow, but it does not provide an explicit user-facing disclosure at the point of transmission. In an agent skill context, this can cause unintentional exfiltration of sensitive local documents if callers assume processing is local.

VirusTotal

53/53 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.