Security audit

ComPDF PDF To Html

Security checks across malware telemetry and agentic risk

Overview

This skill is a narrowly scoped ComPDF PDF-to-HTML helper, with external document upload and local API-key use disclosed and bounded by confirmation instructions.

Install only if you are comfortable having selected PDFs sent to ComPDF for conversion and storing a ComPDF API key in the documented local file. Avoid using the bundled broad reference to perform non-HTML, encryption, decryption, or AI extraction tasks under this skill.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection

Findings (8)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 90% confidence
Finding: The skill instructs the agent to read local API key files and to use network-accessible ComPDF endpoints, but it declares no permissions or capability boundaries. This creates a hidden trust boundary: a caller may invoke the skill believing it is documentation-only, while it can access local secrets and transmit document content externally.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 88% confidence
Finding: The skill is presented as a PDF-to-HTML conversion helper, but the maintainer workflow and referenced behavior include fetching remote documentation, scraping/parsing HTML, and generating local reference snapshots for broader API coverage. This mismatch can conceal extra network and file-write behavior that users and policy systems would not expect, increasing the chance of unauthorized data access, confusing trust decisions, and unintended expansion of the skill's operational scope.

Description-Behavior Mismatch

Medium

Confidence: 96% confidence
Finding: The bundled reference materially exceeds the skill's declared purpose of PDF-to-HTML and documents a broad set of unrelated conversion, editing, security, and AI endpoints. In an agent skill context, this unnecessarily enlarges the action surface and can enable unintended tool use, capability confusion, or prompt-driven invocation of higher-risk APIs that users did not expect.

Context-Inappropriate Capability

High

Confidence: 98% confidence
Finding: Including PDF decryption capability in a PDF-to-HTML skill is high risk because it normalizes and exposes a document-security weakening operation outside the advertised scope. If reachable by the agent, it could be used to strip protections from sensitive documents and facilitate unauthorized downstream access or redistribution.

Context-Inappropriate Capability

Medium

Confidence: 93% confidence
Finding: PDF encryption and permission-control operations are unrelated to PDF-to-HTML conversion and expand the accessible command set beyond user expectations. While not inherently malicious, exposing them in this skill increases the chance of unintended document modification, confusing workflows, or abuse of powerful file-security settings by an autonomous agent.

Context-Inappropriate Capability

Medium

Confidence: 94% confidence
Finding: The AI parsing and extraction material introduces additional data-processing capabilities far beyond PDF-to-HTML, including structured extraction of document contents. In the context of an agent skill, this increases privacy and exfiltration risk because the agent may process more content than necessary or invoke unexpected high-sensitivity extraction features.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The reference explicitly describes uploading user files to external ComPDF endpoints, but the documentation snapshot does not pair this with a clear privacy or off-system transfer warning. In a skill that handles potentially sensitive documents, users may unknowingly transmit confidential content to a third-party processor.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The decryption section describes removing password protection as a normal operation without prominently warning that it reduces or removes document protections. This can lead users or agents to weaken safeguards on sensitive files without appreciating the security consequences.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.