Security audit

Knowledge Importer

Security checks across malware telemetry and agentic risk

Overview

This skill does what it advertises: converts user-selected documents to Markdown, saves them to an Obsidian-style knowledge base, and may upload extracted images to a user-configured image host.

Install only if you are comfortable with converted documents being saved into the configured knowledge-base folder and extracted images being uploaded to the image host you configure. Use a trusted or local image host, avoid importing confidential documents unless you intend those images to be stored there, and test first on non-sensitive files.

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 (5)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 89% confidence
Finding: The skill clearly performs sensitive actions—reading local files, writing into a knowledge base, accessing environment variables, and uploading content/images over the network—yet the manifest declares no permissions or safety boundaries. This creates a transparency and consent failure: users or orchestrators may invoke the skill without understanding that local documents can be persisted and externally transmitted.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The README explicitly promotes uploading extracted document images to an external image host and generating external URLs, but it does not prominently warn that embedded document content may leave the local system. In a knowledge-import workflow, imported files may contain sensitive screenshots, diagrams, or confidential material, so silent or poorly disclosed exfiltration to a remote server creates a real privacy and data-leak risk.

Vague Triggers

Medium

Confidence: 81% confidence
Finding: The trigger description is broad enough to match many generic document-handling requests, which increases the chance the skill runs in contexts where the user did not specifically intend import, conversion, storage, or image upload. In this skill's context, mis-triggering is more dangerous because invocation can lead to file ingestion, persistent writes, and possible external exfiltration of embedded images.

Missing User Warnings

High

Confidence: 95% confidence
Finding: The documentation states that converted content will be written into an Obsidian knowledge base and that images may be uploaded to an external image bed, but it does not prominently warn about the privacy, retention, and data-transfer risks. Because imported documents may contain sensitive business or personal information, silent persistence and external upload materially increase the risk of unintended disclosure.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The script automatically uploads extracted images from user documents to a configurable external server, which can exfiltrate sensitive document content outside the local knowledge base. In a document-import skill, users may reasonably expect conversion and local storage, but not silent network transmission of embedded images, making the skill context more dangerous because imported files often contain confidential material.

VirusTotal

67/67 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.